Add a basic file system module for virtio-fs. This does not yet contain
shared data support between host and guest or metadata coherency speedups.
However it is already significantly faster than virtio-9p.
Design Overview
===============
With the goal of designing something with better performance and local file
system semantics, a bunch of ideas were proposed.
- Use fuse protocol (instead of 9p) for communication between guest and
host. Guest kernel will be fuse client and a fuse server will run on
host to serve the requests.
- For data access inside guest, mmap portion of file in QEMU address space
and guest accesses this memory using dax. That way guest page cache is
bypassed and there is only one copy of data (on host). This will also
enable mmap(MAP_SHARED) between guests.
- For metadata coherency, there is a shared memory region which contains
version number associated with metadata and any guest changing metadata
updates version number and other guests refresh metadata on next access.
This is yet to be implemented.
How virtio-fs differs from existing approaches
==============================================
The unique idea behind virtio-fs is to take advantage of the co-location of
the virtual machine and hypervisor to avoid communication (vmexits).
DAX allows file contents to be accessed without communication with the
hypervisor. The shared memory region for metadata avoids communication in
the common case where metadata is unchanged.
By replacing expensive communication with cheaper shared memory accesses,
we expect to achieve better performance than approaches based on network
file system protocols. In addition, this also makes it easier to achieve
local file system semantics (coherency).
These techniques are not applicable to network file system protocols since
the communications channel is bypassed by taking advantage of shared memory
on a local machine. This is why we decided to build virtio-fs rather than
focus on 9P or NFS.
Caching Modes
=============
Like virtio-9p, different caching modes are supported which determine the
coherency level as well. The “cache=FOO” and “writeback” options control
the level of coherence between the guest and host filesystems.
- cache=none
metadata, data and pathname lookup are not cached in guest. They are
always fetched from host and any changes are immediately pushed to host.
- cache=always
metadata, data and pathname lookup are cached in guest and never expire.
- cache=auto
metadata and pathname lookup cache expires after a configured amount of
time (default is 1 second). Data is cached while the file is open
(close to open consistency).
- writeback/no_writeback
These options control the writeback strategy. If writeback is disabled,
then normal writes will immediately be synchronized with the host fs.
If writeback is enabled, then writes may be cached in the guest until
the file is closed or an fsync(2) performed. This option has no effect
on mmap-ed writes or writes going through the DAX mechanism.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Vivek Goyal <vgoyal@redhat.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
virtio-fs does not support aborting requests which are being
processed. That is requests which have been sent to fuse daemon on host.
Signed-off-by: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Don't hold onto dentry in lru list if need to re-lookup it anyway at next
access. Only do this if explicitly enabled, otherwise it could result in
performance regression.
More advanced version of this patch would periodically flush out dentries
from the lru which have gone stale.
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
As of now fuse_dev_alloc() both allocates a fuse device and installs it in
fuse_conn list. fuse_dev_alloc() can fail if fuse_device allocation fails.
virtio-fs needs to initialize multiple fuse devices (one per virtio queue).
It initializes one fuse device as part of call to fuse_fill_super_common()
and rest of the devices are allocated and installed after that.
But, we can't afford to fail after calling fuse_fill_super_common() as we
don't have a way to undo all the actions done by fuse_fill_super_common().
So to avoid failures after the call to fuse_fill_super_common(),
pre-allocate all fuse devices early and install them into fuse connection
later.
This patch provides two separate helpers for fuse device allocation and
fuse device installation in fuse_conn.
Signed-off-by: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
The /dev/fuse device uses fiq->waitq and fasync to signal that requests are
available. These mechanisms do not apply to virtio-fs. This patch
introduces callbacks so alternative behavior can be used.
Note that queue_interrupt() changes along these lines:
spin_lock(&fiq->waitq.lock);
wake_up_locked(&fiq->waitq);
+ kill_fasync(&fiq->fasync, SIGIO, POLL_IN);
spin_unlock(&fiq->waitq.lock);
- kill_fasync(&fiq->fasync, SIGIO, POLL_IN);
Since queue_request() and queue_forget() also call kill_fasync() inside
the spinlock this should be safe.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
fuse_fill_super() includes code to process the fd= option and link the
struct fuse_dev to the fd's struct file. In virtio-fs there is no file
descriptor because /dev/fuse is not used.
This patch extracts fuse_fill_super_common() so that both classic fuse and
virtio-fs can share the code to initialize a mount.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
File systems like virtio-fs need to do not have to play directly with
forget list data structures. There is a helper function use that instead.
Rename dequeue_forget() to fuse_dequeue_forget() and export it so that
stacked filesystems can use it.
Signed-off-by: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
virtio-fs will need unique IDs for FORGET requests from outside
fs/fuse/dev.c. Make the symbol visible.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
This will be used by virtio-fs to send init request to fuse server after
initialization of virt queues.
Signed-off-by: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
virtio-fs will need to query the length of fuse_arg lists. Make the symbol
visible.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
virtio-fs will need to complete requests from outside fs/fuse/dev.c. Make
the symbol visible.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
The page array pointers are also duplicated across fuse_args_pages and
fuse_req. Get rid of the fuse_req ones.
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
No need to duplicate the argument arrays in fuse_req, so just dereference
req->args instead of copying to the fuse_req internal ones.
This allows further cleanup of the fuse_req structure.
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
All requests are now sent with one of the fuse_simple_... helpers. Get rid
of the old api from the fuse internal header.
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Since we cannot reserve the request structure up-front, make sure that the
request allocation doesn't fail using __GFP_NOFAIL.
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Derive fuse_writepage_args from fuse_io_args.
Sending the request is tricky since it was done with fi->lock held, hence
we must either use atomic allocation or release the lock. Both are
possible so try atomic first and if it fails, release the lock and do the
regular allocation with GFP_NOFS and __GFP_NOFAIL. Both flags are
necessary for correct operation.
Move the page realloc function from dev.c to file.c and convert to using
fuse_writepage_args.
The last caller of fuse_write_fill() is gone, so get rid of it.
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
The old fuse_read_fill() helper can be deleted, now that the last user is
gone.
The fuse_io_args struct is moved to fuse_i.h so it can be shared between
readdir/read code.
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Create a helper named fuse_simple_background() that is similar to
fuse_simple_request(). Unlike the latter, it returns immediately and calls
the supplied 'end' callback when the reply is received.
The supplied 'args' pointer is stored in 'fuse_req' which allows the
callback to interpret the output arguments decoded from the reply.
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
fuse_req_pages_alloc() is moved to file.c, since its internal use by the
device code will eventually be removed.
Rename to fuse_pages_alloc() to signify that it's not only usable for
fuse_req page array.
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Derive fuse_args_pages from fuse_args. This is used to handle requests
which use pages for input or output. The related flags are added to
fuse_args.
New FR_ALLOC_PAGES flags is added to indicate whether the page arrays in
fuse_req need to be freed by fuse_put_request() or not.
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
We can use the "force" flag to make sure the DESTROY request is always sent
to userspace. So no need to keep it allocated during the lifetime of the
filesystem.
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
In some cases it makes no sense to set pid/uid/gid fields in the request
header. Allow fuse_simple_background() to omit these. This is only
required in the "force" case, so for now just WARN if set otherwise.
Fold fuse_get_req_nofail_nopages() into its only caller. Comment is
obsolete anyway.
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
This will be used by fuse_force_forget().
We can expand fuse_request_send() into fuse_simple_request(). The
FR_WAITING bit has already been set, no need to check.
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Add 'force' to fuse_args and use fuse_get_req_nofail_nopages() to allocate
the request in that case.
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Instead of complex games with a reserved request, just use __GFP_NOFAIL.
Both calers (flush, readdir) guarantee that connection was already
initialized, so no need to wait for fc->initialized.
Also remove unneeded clearing of FR_BACKGROUND flag.
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
...to make future expansion simpler. The hiearachical structure is a
historical thing that does not serve any practical purpose.
The generated code is excatly the same before and after the patch.
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
When IOCB_CMD_POLL is used on the FUSE device, aio_poll() disables IRQs
and takes kioctx::ctx_lock, then fuse_iqueue::waitq.lock.
This may have to wait for fuse_iqueue::waitq.lock to be released by one
of many places that take it with IRQs enabled. Since the IRQ handler
may take kioctx::ctx_lock, lockdep reports that a deadlock is possible.
Fix it by protecting the state of struct fuse_iqueue with a separate
spinlock, and only accessing fuse_iqueue::waitq using the versions of
the waitqueue functions which do IRQ-safe locking internally.
Reproducer:
#include <fcntl.h>
#include <stdio.h>
#include <sys/mount.h>
#include <sys/stat.h>
#include <sys/syscall.h>
#include <unistd.h>
#include <linux/aio_abi.h>
int main()
{
char opts[128];
int fd = open("/dev/fuse", O_RDWR);
aio_context_t ctx = 0;
struct iocb cb = { .aio_lio_opcode = IOCB_CMD_POLL, .aio_fildes = fd };
struct iocb *cbp = &cb;
sprintf(opts, "fd=%d,rootmode=040000,user_id=0,group_id=0", fd);
mkdir("mnt", 0700);
mount("foo", "mnt", "fuse", 0, opts);
syscall(__NR_io_setup, 1, &ctx);
syscall(__NR_io_submit, ctx, 1, &cbp);
}
Beginning of lockdep output:
=====================================================
WARNING: SOFTIRQ-safe -> SOFTIRQ-unsafe lock order detected
5.3.0-rc5 #9 Not tainted
-----------------------------------------------------
syz_fuse/135 [HC0[0]:SC0[0]:HE0:SE1] is trying to acquire:
000000003590ceda (&fiq->waitq){+.+.}, at: spin_lock include/linux/spinlock.h:338 [inline]
000000003590ceda (&fiq->waitq){+.+.}, at: aio_poll fs/aio.c:1751 [inline]
000000003590ceda (&fiq->waitq){+.+.}, at: __io_submit_one.constprop.0+0x203/0x5b0 fs/aio.c:1825
and this task is already holding:
0000000075037284 (&(&ctx->ctx_lock)->rlock){..-.}, at: spin_lock_irq include/linux/spinlock.h:363 [inline]
0000000075037284 (&(&ctx->ctx_lock)->rlock){..-.}, at: aio_poll fs/aio.c:1749 [inline]
0000000075037284 (&(&ctx->ctx_lock)->rlock){..-.}, at: __io_submit_one.constprop.0+0x1f4/0x5b0 fs/aio.c:1825
which would create a new lock dependency:
(&(&ctx->ctx_lock)->rlock){..-.} -> (&fiq->waitq){+.+.}
but this new dependency connects a SOFTIRQ-irq-safe lock:
(&(&ctx->ctx_lock)->rlock){..-.}
[...]
Reported-by: syzbot+af05535bb79520f95431@syzkaller.appspotmail.com
Reported-by: syzbot+d86c4426a01f60feddc7@syzkaller.appspotmail.com
Fixes: bfe4037e72 ("aio: implement IOCB_CMD_POLL")
Cc: <stable@vger.kernel.org> # v4.19+
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
On networked filesystems file data can be changed externally. FUSE
provides notification messages for filesystem to inform kernel that
metadata or data region of a file needs to be invalidated in local page
cache. That provides the basis for filesystem implementations to invalidate
kernel cache explicitly based on observed filesystem-specific events.
FUSE has also "automatic" invalidation mode(*) when the kernel
automatically invalidates data cache of a file if it sees mtime change. It
also automatically invalidates whole data cache of a file if it sees file
size being changed.
The automatic mode has corresponding capability - FUSE_AUTO_INVAL_DATA.
However, due to probably historical reason, that capability controls only
whether mtime change should be resulting in automatic invalidation or
not. A change in file size always results in invalidating whole data cache
of a file irregardless of whether FUSE_AUTO_INVAL_DATA was negotiated(+).
The filesystem I write[1] represents data arrays stored in networked
database as local files suitable for mmap. It is read-only filesystem -
changes to data are committed externally via database interfaces and the
filesystem only glues data into contiguous file streams suitable for mmap
and traditional array processing. The files are big - starting from
hundreds gigabytes and more. The files change regularly, and frequently by
data being appended to their end. The size of files thus changes
frequently.
If a file was accessed locally and some part of its data got into page
cache, we want that data to stay cached unless there is memory pressure, or
unless corresponding part of the file was actually changed. However current
FUSE behaviour - when it sees file size change - is to invalidate the whole
file. The data cache of the file is thus completely lost even on small size
change, and despite that the filesystem server is careful to accurately
translate database changes into FUSE invalidation messages to kernel.
Let's fix it: if a filesystem, through new FUSE_EXPLICIT_INVAL_DATA
capability, indicates to kernel that it is fully responsible for data cache
invalidation, then the kernel won't invalidate files data cache on size
change and only truncate that cache to new size in case the size decreased.
(*) see 72d0d248ca "fuse: add FUSE_AUTO_INVAL_DATA init flag",
eed2179efe "fuse: invalidate inode mapping if mtime changes"
(+) in writeback mode the kernel does not invalidate data cache on file
size change, but neither it allows the filesystem to set the size due to
external event (see 8373200b12 "fuse: Trust kernel i_size only")
[1] https://lab.nexedi.com/kirr/wendelin.core/blob/a50f1d9f/wcfs/wcfs.go#L20
Signed-off-by: Kirill Smelkov <kirr@nexedi.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Functions, like pr_err, are a more modern variant of printing compared to
printk. They could be used to denoise sources by using needed level in
the print function name, and by automatically inserting per-driver /
function / ... print prefix as defined by pr_fmt macro. pr_* are also
said to be used in Documentation/process/coding-style.rst and more
recent code - for example overlayfs - uses them instead of printk.
Convert CUSE and FUSE to use the new pr_* functions.
CUSE output stays completely unchanged, while FUSE output is amended a
bit for "trying to steal weird page" warning - the second line now comes
also with "fuse:" prefix. I hope it is ok.
Suggested-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Kirill Smelkov <kirr@nexedi.com>
Reviewed-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Allow filesystems to return ENOSYS from opendir, preventing the kernel from
sending opendir and releasedir messages in the future. This avoids
userspace transitions when filesystems don't need to keep track of state
per directory handle.
A new capability flag, FUSE_NO_OPENDIR_SUPPORT, parallels
FUSE_NO_OPEN_SUPPORT, indicating the new semantics for returning ENOSYS
from opendir.
Signed-off-by: Chad Austin <chadaustin@fb.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
The only caller that needs fc->aborted set is fuse_conn_abort_write().
Setting fc->aborted is now racy (fuse_abort_conn() may already be in
progress or finished) but there's no reason to care.
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
This is rather natural action after previous patches, and it just decreases
load of fc->lock.
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
To minimize contention of fc->lock, this patch introduces a new spinlock
for protection fuse_inode metadata:
fuse_inode:
writectr
writepages
write_files
queued_writes
attr_version
inode:
i_size
i_nlink
i_mtime
i_ctime
Also, it protects the fields changed in fuse_change_attributes_common()
(too many to list).
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
This patch makes fc->attr_version of atomic64_t type, so fc->lock won't be
needed to read or modify it anymore.
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Here is preparation for next patches, which introduce new fi->lock for
protection of ff->write_entry linked into fi->write_files.
This patch just passes new argument to the function.
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
When FUSE_OPEN returns ENOSYS, the no_open bit is set on the connection.
Because the FUSE_RELEASE and FUSE_RELEASEDIR paths share code, this
incorrectly caused the FUSE_RELEASEDIR request to be dropped and never sent
to userspace.
Pass an isdir bool to distinguish between FUSE_RELEASE and FUSE_RELEASEDIR
inside of fuse_file_put.
Fixes: 7678ac5061 ("fuse: support clients that don't implement 'open'")
Cc: <stable@vger.kernel.org> # v3.14
Signed-off-by: Chad Austin <chadaustin@fb.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Commit ab2257e994 ("fuse: reduce size of struct fuse_inode") moved parts
of fields related to writeback on regular file and to directory caching
into a union. However fuse_fsync_common() called from fuse_dir_fsync()
touches some writeback related fields, resulting in a crash.
Move writeback related parts from fuse_fsync_common() to fuse_fysnc().
Reported-by: Brett Girton <btgirton@gmail.com>
Tested-by: Brett Girton <btgirton@gmail.com>
Fixes: ab2257e994 ("fuse: reduce size of struct fuse_inode")
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
FUSE file reads are cached in the page cache, but symlink reads are
not. This patch enables FUSE READLINK operations to be cached which
can improve performance of some FUSE workloads.
In particular, I'm working on a FUSE filesystem for access to source
code and discovered that about a 10% improvement to build times is
achieved with this patch (there are a lot of symlinks in the source
tree).
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
This patch adds the infrastructure for more fine grained attribute
invalidation. Currently only 'atime' is invalidated separately.
The use of this infrastructure is extended to the statx(2) interface, which
for now means that if only 'atime' is invalid and STATX_ATIME is not
specified in the mask argument, then no GETATTR request will be generated.
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Writeback caching currently allocates requests with the maximum number of
possible pages, while the actual number of pages per request depends on a
couple of factors that cannot be determined when the request is allocated
(whether page is already under writeback, whether page is contiguous with
previous pages already added to a request).
This patch allows such requests to start with no page allocation (all pages
inline) and grow the page array on demand.
If the max_pages tunable remains the default value, then this will mean
just one allocation that is the same size as before. If the tunable is
larger, then this adds at most 3 additional memory allocations (which is
generously compensated by the improved performance from the larger
request).
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Replace FUSE_MAX_PAGES_PER_REQ with the configurable parameter max_pages to
improve performance.
Old RFC with detailed description of the problem and many fixes by Mitsuo
Hayasaka (mitsuo.hayasaka.hu@hitachi.com):
- https://lkml.org/lkml/2012/7/5/136
We've encountered performance degradation and fixed it on a big and complex
virtual environment.
Environment to reproduce degradation and improvement:
1. Add lag to user mode FUSE
Add nanosleep(&(struct timespec){ 0, 1000 }, NULL); to xmp_write_buf in
passthrough_fh.c
2. patch UM fuse with configurable max_pages parameter. The patch will be
provided latter.
3. run test script and perform test on tmpfs
fuse_test()
{
cd /tmp
mkdir -p fusemnt
passthrough_fh -o max_pages=$1 /tmp/fusemnt
grep fuse /proc/self/mounts
dd conv=fdatasync oflag=dsync if=/dev/zero of=fusemnt/tmp/tmp \
count=1K bs=1M 2>&1 | grep -v records
rm fusemnt/tmp/tmp
killall passthrough_fh
}
Test results:
passthrough_fh /tmp/fusemnt fuse.passthrough_fh \
rw,nosuid,nodev,relatime,user_id=0,group_id=0 0 0
1073741824 bytes (1.1 GB) copied, 1.73867 s, 618 MB/s
passthrough_fh /tmp/fusemnt fuse.passthrough_fh \
rw,nosuid,nodev,relatime,user_id=0,group_id=0,max_pages=256 0 0
1073741824 bytes (1.1 GB) copied, 1.15643 s, 928 MB/s
Obviously with bigger lag the difference between 'before' and 'after'
will be more significant.
Mitsuo Hayasaka, in 2012 (https://lkml.org/lkml/2012/7/5/136),
observed improvement from 400-550 to 520-740.
Signed-off-by: Constantine Shulyupin <const@MakeLinux.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Do this by grouping fields used for cached writes and putting them into a
union with fileds used for cached readdir (with obviously no overlap, since
we don't have hybrid objects).
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Use the internal iversion counter to make sure modifications of the
directory through this filesystem are not missed by the mtime check (due to
mtime granularity).
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Store the modification time of the directory in the cache, obtained before
starting to fill the cache.
When reading the cache, verify that the directory hasn't changed, by
checking if current modification time is the same as the one stored in the
cache.
This only needs to be done when the current file position is at the
beginning of the directory, as mandated by POSIX.
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Stefan Hajnoczi