From: "Darrick J. Wong" <djwong@kernel.org>
To: tytso@mit.edu
Cc: amir73il@gmail.com, John@groves.net, bernd@bsbernd.com,
linux-fsdevel@vger.kernel.org, linux-ext4@vger.kernel.org,
miklos@szeredi.hu, amir73il@gmail.com, joannelkoong@gmail.com,
neal@gompa.dev
Subject: [PATCHSET RFC v4 1/6] fuse4fs: fork a low level fuse server
Date: Wed, 20 Aug 2025 17:49:22 -0700 [thread overview]
Message-ID: <175573712721.20753.5223489399594191991.stgit@frogsfrogsfrogs> (raw)
In-Reply-To: <20250821003720.GA4194186@frogsfrogsfrogs>
Hi all,
Whilst developing the fuse2fs+iomap prototype, I discovered a
fundamental design limitation of the upper-level libfuse API: hardlinks.
The upper level fuse library really wants to communicate with the fuse
server with file paths, instead of using inode numbers. This works
great for filesystems that don't have inodes, create files dynamically
at runtime, or lack stable inode numbers.
Unfortunately, the libfuse path abstraction assigns a unique nodeid to
every child file in the entire filesystem, without regard to hard links.
In other words, a hardlinked regular file may have one ondisk inode
number but multiple kernel inodes. For classic fuse2fs this isn't a
problem because all file access goes through the fuse server and the big
library lock protects us from corruption.
For fuse2fs + iomap this is a disaster because we rely on the kernel to
coordinate access to inodes. For hardlinked files, we *require* that
there only be one in-kernel inode for each ondisk inode.
The path based mechanism is also very inefficient for fuse2fs. Every
time a file is accessed, the upper level libfuse passes a new nodeid to
the kernel, and on every file access the kernel passes that same nodeid
back to libfuse. libfuse then walks its internal directory entry cache
to construct a path string for that nodeid and hands it to fuse2fs.
fuse2fs then walks the ondisk directory structure to find the ext2 inode
number. Every time.
Create a new fuse4fs server from fuse2fs that uses the lowlevel fuse
API. This affords us direct control over nodeids and eliminates the
path wrangling. Hardlinks can be supported when iomap is turned on,
and metadata-heavy workloads run twice as fast.
If you're going to start using this code, I strongly recommend pulling
from my git trees, which are linked below.
Comments and questions are, as always, welcome.
e2fsprogs git tree:
https://git.kernel.org/cgit/linux/kernel/git/djwong/e2fsprogs.git/log/?h=fuse4fs-fork
---
Commits in this patchset:
* fuse2fs: port fuse2fs to lowlevel libfuse API
* fuse4fs: drop fuse 2.x support code
* fuse4fs: namespace some helpers
* fuse4fs: convert to low level API
* libsupport: port the kernel list.h to libsupport
* libsupport: add a cache
* cache: disable debugging
* cache: use modern list iterator macros
* cache: embed struct cache in the owner
* cache: pass cache pointer to callbacks
* cache: pass a private data pointer through cache_walk
* cache: add a helper to grab a new refcount for a cache_node
* cache: return results of a cache flush
* cache: add a "get only if incore" flag to cache_node_get
* cache: support gradual expansion
* cache: implement automatic shrinking
* fuse4fs: add cache to track open files
* fuse4fs: use the orphaned inode list
* fuse4fs: implement FUSE_TMPFILE
* fuse4fs: create incore reverse orphan list
---
lib/ext2fs/jfs_compat.h | 2
lib/ext2fs/kernel-list.h | 111 -
lib/support/cache.h | 177 +
lib/support/list.h | 901 +++++++
lib/support/xbitops.h | 128 +
configure | 50
configure.ac | 31
debugfs/Makefile.in | 12
e2fsck/Makefile.in | 56
lib/config.h.in | 3
lib/e2p/Makefile.in | 4
lib/ext2fs/Makefile.in | 14
lib/support/Makefile.in | 8
lib/support/cache.c | 853 ++++++
misc/Makefile.in | 35
misc/fuse4fs.c | 6098 ++++++++++++++++++++++++++++++++++++++++++++++
misc/tune2fs.c | 4
17 files changed, 8319 insertions(+), 168 deletions(-)
delete mode 100644 lib/ext2fs/kernel-list.h
create mode 100644 lib/support/cache.h
create mode 100644 lib/support/list.h
create mode 100644 lib/support/xbitops.h
create mode 100644 lib/support/cache.c
create mode 100644 misc/fuse4fs.c
next prev parent reply other threads:[~2025-08-21 0:49 UTC|newest]
Thread overview: 72+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-08-21 0:37 [RFC v4] fuse: use fs-iomap for better performance so we can containerize ext4 Darrick J. Wong
2025-08-21 0:49 ` Darrick J. Wong [this message]
2025-08-21 1:08 ` [PATCH 01/20] fuse2fs: port fuse2fs to lowlevel libfuse API Darrick J. Wong
2025-08-21 1:08 ` [PATCH 02/20] fuse4fs: drop fuse 2.x support code Darrick J. Wong
2025-08-21 1:08 ` [PATCH 03/20] fuse4fs: namespace some helpers Darrick J. Wong
2025-08-21 1:08 ` [PATCH 04/20] fuse4fs: convert to low level API Darrick J. Wong
2025-08-21 1:09 ` [PATCH 05/20] libsupport: port the kernel list.h to libsupport Darrick J. Wong
2025-08-21 1:09 ` [PATCH 06/20] libsupport: add a cache Darrick J. Wong
2025-08-21 1:09 ` [PATCH 07/20] cache: disable debugging Darrick J. Wong
2025-08-21 1:09 ` [PATCH 08/20] cache: use modern list iterator macros Darrick J. Wong
2025-08-21 1:10 ` [PATCH 09/20] cache: embed struct cache in the owner Darrick J. Wong
2025-08-21 1:10 ` [PATCH 10/20] cache: pass cache pointer to callbacks Darrick J. Wong
2025-08-21 1:10 ` [PATCH 11/20] cache: pass a private data pointer through cache_walk Darrick J. Wong
2025-08-21 1:11 ` [PATCH 12/20] cache: add a helper to grab a new refcount for a cache_node Darrick J. Wong
2025-08-21 1:11 ` [PATCH 13/20] cache: return results of a cache flush Darrick J. Wong
2025-08-21 1:11 ` [PATCH 14/20] cache: add a "get only if incore" flag to cache_node_get Darrick J. Wong
2025-08-21 1:11 ` [PATCH 15/20] cache: support gradual expansion Darrick J. Wong
2025-08-21 1:12 ` [PATCH 16/20] cache: implement automatic shrinking Darrick J. Wong
2025-08-21 1:12 ` [PATCH 17/20] fuse4fs: add cache to track open files Darrick J. Wong
2025-08-21 1:12 ` [PATCH 18/20] fuse4fs: use the orphaned inode list Darrick J. Wong
2025-08-21 1:12 ` [PATCH 19/20] fuse4fs: implement FUSE_TMPFILE Darrick J. Wong
2025-08-21 1:13 ` [PATCH 20/20] fuse4fs: create incore reverse orphan list Darrick J. Wong
2025-08-21 0:49 ` [PATCHSET RFC v4 2/6] libext2fs: refactoring for fuse2fs iomap support Darrick J. Wong
2025-08-21 1:13 ` [PATCH 01/10] libext2fs: make it possible to extract the fd from an IO manager Darrick J. Wong
2025-08-21 1:13 ` [PATCH 02/10] libext2fs: always fsync the device when flushing the cache Darrick J. Wong
2025-08-21 1:13 ` [PATCH 03/10] libext2fs: always fsync the device when closing the unix IO manager Darrick J. Wong
2025-08-21 1:14 ` [PATCH 04/10] libext2fs: only fsync the unix fd if we wrote to the device Darrick J. Wong
2025-08-21 1:14 ` [PATCH 05/10] libext2fs: invalidate cached blocks when freeing them Darrick J. Wong
2025-08-21 1:14 ` [PATCH 06/10] libext2fs: only flush affected blocks in unix_write_byte Darrick J. Wong
2025-08-21 1:14 ` [PATCH 07/10] libext2fs: allow unix_write_byte when the write would be aligned Darrick J. Wong
2025-08-21 1:15 ` [PATCH 08/10] libext2fs: allow clients to ask to write full superblocks Darrick J. Wong
2025-08-21 1:15 ` [PATCH 09/10] libext2fs: allow callers to disallow I/O to file data blocks Darrick J. Wong
2025-08-21 1:15 ` [PATCH 10/10] libext2fs: add posix advisory locking to the unix IO manager Darrick J. Wong
2025-08-21 0:49 ` [PATCHSET RFC v4 3/6] fuse2fs: use fuse iomap data paths for better file I/O performance Darrick J. Wong
2025-08-21 1:15 ` [PATCH 01/19] fuse2fs: implement bare minimum iomap for file mapping reporting Darrick J. Wong
2025-08-21 1:16 ` [PATCH 02/19] fuse2fs: add iomap= mount option Darrick J. Wong
2025-08-21 1:16 ` [PATCH 03/19] fuse2fs: implement iomap configuration Darrick J. Wong
2025-08-21 1:16 ` [PATCH 04/19] fuse2fs: register block devices for use with iomap Darrick J. Wong
2025-08-21 1:17 ` [PATCH 05/19] fuse2fs: implement directio file reads Darrick J. Wong
2025-08-21 1:17 ` [PATCH 06/19] fuse2fs: add extent dump function for debugging Darrick J. Wong
2025-08-21 1:17 ` [PATCH 07/19] fuse2fs: implement direct write support Darrick J. Wong
2025-08-21 1:17 ` [PATCH 08/19] fuse2fs: turn on iomap for pagecache IO Darrick J. Wong
2025-08-21 1:18 ` [PATCH 09/19] fuse2fs: don't zero bytes in punch hole Darrick J. Wong
2025-08-21 1:18 ` [PATCH 10/19] fuse2fs: don't do file data block IO when iomap is enabled Darrick J. Wong
2025-08-21 1:18 ` [PATCH 11/19] fuse2fs: avoid fuseblk mode if fuse-iomap support is likely Darrick J. Wong
2025-08-21 1:18 ` [PATCH 12/19] fuse2fs: enable file IO to inline data files Darrick J. Wong
2025-08-21 1:19 ` [PATCH 13/19] fuse2fs: set iomap-related inode flags Darrick J. Wong
2025-08-21 1:19 ` [PATCH 14/19] fuse2fs: add strictatime/lazytime mount options Darrick J. Wong
2025-08-21 1:19 ` [PATCH 15/19] fuse2fs: configure block device block size Darrick J. Wong
2025-08-21 1:19 ` [PATCH 16/19] fuse4fs: don't use inode number translation when possible Darrick J. Wong
2025-08-21 1:20 ` [PATCH 17/19] fuse4fs: separate invalidation Darrick J. Wong
2025-08-21 1:20 ` [PATCH 18/19] fuse2fs: implement statx Darrick J. Wong
2025-08-21 1:20 ` [PATCH 19/19] fuse2fs: enable atomic writes Darrick J. Wong
2025-08-21 0:50 ` [PATCHSET RFC v4 4/6] fuse2fs: use fuse iomap data paths for better file I/O performance Darrick J. Wong
2025-08-21 1:20 ` [PATCH 1/2] fuse2fs: enable caching of iomaps Darrick J. Wong
2025-08-21 1:21 ` [PATCH 2/2] fuse2fs: be smarter about caching iomaps Darrick J. Wong
2025-08-21 0:50 ` [PATCHSET RFC v4 5/6] fuse2fs: handle timestamps and ACLs correctly when iomap is enabled Darrick J. Wong
2025-08-21 1:21 ` [PATCH 1/8] fuse2fs: skip permission checking on utimens " Darrick J. Wong
2025-08-21 1:21 ` [PATCH 2/8] fuse2fs: let the kernel tell us about acl/mode updates Darrick J. Wong
2025-08-21 1:21 ` [PATCH 3/8] fuse2fs: better debugging for file mode updates Darrick J. Wong
2025-08-21 1:22 ` [PATCH 4/8] fuse2fs: debug timestamp updates Darrick J. Wong
2025-08-21 1:22 ` [PATCH 5/8] fuse2fs: use coarse timestamps for iomap mode Darrick J. Wong
2025-08-21 1:22 ` [PATCH 6/8] fuse2fs: add tracing for retrieving timestamps Darrick J. Wong
2025-08-21 1:23 ` [PATCH 7/8] fuse2fs: enable syncfs Darrick J. Wong
2025-08-21 1:23 ` [PATCH 8/8] fuse2fs: skip the gdt write in op_destroy if syncfs is working Darrick J. Wong
2025-08-21 0:50 ` [PATCHSET RFC v4 6/6] fuse2fs: improve block and inode caching Darrick J. Wong
2025-08-21 1:23 ` [PATCH 1/6] libsupport: add caching IO manager Darrick J. Wong
2025-08-21 1:23 ` [PATCH 2/6] iocache: add the actual buffer cache Darrick J. Wong
2025-08-21 1:24 ` [PATCH 3/6] iocache: bump buffer mru priority every 50 accesses Darrick J. Wong
2025-08-21 1:24 ` [PATCH 4/6] fuse2fs: enable caching IO manager Darrick J. Wong
2025-08-21 1:24 ` [PATCH 5/6] fuse2fs: increase inode cache size Darrick J. Wong
2025-08-21 1:24 ` [PATCH 6/6] libext2fs: improve caching for inodes Darrick J. Wong
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=175573712721.20753.5223489399594191991.stgit@frogsfrogsfrogs \
--to=djwong@kernel.org \
--cc=John@groves.net \
--cc=amir73il@gmail.com \
--cc=bernd@bsbernd.com \
--cc=joannelkoong@gmail.com \
--cc=linux-ext4@vger.kernel.org \
--cc=linux-fsdevel@vger.kernel.org \
--cc=miklos@szeredi.hu \
--cc=neal@gompa.dev \
--cc=tytso@mit.edu \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox