| Age | Commit message (Collapse) | Author |
|---|
|
Because there is no inode structure stored in the sandbox, inotify watches
must be held on the dentry. This would be an issue in the presence of hard
links, where multiple dentries would need to share the same set of watches,
but in VFS2, we do not support the internal creation of hard links on gofer
fs. As a result, we make the assumption that every dentry corresponds to a
unique inode.
Furthermore, dentries can be cached and then evicted, even if the underlying
file has not be deleted. We must prevent this from occurring if there are any
watches that would be lost. Note that if the dentry was deleted or invalidated
(d.vfsd.IsDead()), we should still destroy it along with its watches.
Additionally, when a dentry’s last watch is removed, we cache it if it also
has zero references. This way, the dentry can eventually be evicted from
memory if it is no longer needed. This is accomplished with a new dentry
method, OnZeroWatches(), which is called by Inotify.RmWatch and
Inotify.Release. Note that it must be called after all inotify locks are
released to avoid violating lock order. Stress tests are added to make sure
that inotify operations don't deadlock with gofer.OnZeroWatches.
Updates #1479.
PiperOrigin-RevId: 317958034
|
|
Limited to tmpfs. Inotify support in other filesystem implementations to
follow.
Updates #1479
PiperOrigin-RevId: 313828648
|
|
Support in other filesystem impls is still needed. Unlike in Linux and vfs1, we
need to plumb inotify down to each filesystem implementation in order to keep
track of links/inode structures properly.
IN_EXCL_UNLINK still needs to be implemented, as well as a few inotify hooks
that are not present in either vfs1 or vfs2. Those will be addressed in
subsequent changes.
Updates #1479.
PiperOrigin-RevId: 313781995
|
|
This change:
- Drastically simplifies the synchronization model: filesystem structure is
both implementation-defined and implementation-synchronized.
- Allows implementations of vfs.DentryImpl to use implementation-specific
dentry types, reducing casts during path traversal.
- Doesn't require dentries representing non-directory files to waste space on a
map of children.
- Allows dentry revalidation and mount lookup to be correctly ordered (fixed
FIXME in fsimpl/gofer/filesystem.go).
- Removes the need to have two separate maps in gofer.dentry
(dentry.vfsd.children and dentry.negativeChildren) for positive and negative
lookups respectively.
//pkg/sentry/fsimpl/tmpfs/benchmark_test.go:
name old time/op new time/op delta
VFS2TmpfsStat/1-112 172ns ± 4% 165ns ± 3% -4.08% (p=0.002 n=9+9)
VFS2TmpfsStat/2-112 199ns ± 3% 195ns ±10% ~ (p=0.132 n=8+9)
VFS2TmpfsStat/3-112 230ns ± 2% 216ns ± 2% -6.15% (p=0.000 n=8+8)
VFS2TmpfsStat/8-112 390ns ± 2% 358ns ± 4% -8.33% (p=0.000 n=9+8)
VFS2TmpfsStat/64-112 2.20µs ± 3% 2.01µs ± 3% -8.48% (p=0.000 n=10+8)
VFS2TmpfsStat/100-112 3.42µs ± 9% 3.08µs ± 2% -9.82% (p=0.000 n=9+8)
VFS2TmpfsMountStat/1-112 278ns ± 1% 286ns ±15% ~ (p=0.712 n=8+10)
VFS2TmpfsMountStat/2-112 311ns ± 4% 298ns ± 2% -4.27% (p=0.000 n=9+8)
VFS2TmpfsMountStat/3-112 339ns ± 3% 330ns ± 9% ~ (p=0.070 n=8+9)
VFS2TmpfsMountStat/8-112 503ns ± 3% 466ns ± 3% -7.38% (p=0.000 n=8+8)
VFS2TmpfsMountStat/64-112 2.53µs ±16% 2.17µs ± 7% -14.19% (p=0.000 n=10+9)
VFS2TmpfsMountStat/100-112 3.60µs ± 4% 3.30µs ± 8% -8.33% (p=0.001 n=8+9)
Updates #1035
PiperOrigin-RevId: 307655892
|
|
This saves one pointer dereference per VFS access.
Updates #1623
PiperOrigin-RevId: 295216176
|
|
* Rename syncutil to sync.
* Add aliases to sync types.
* Replace existing usage of standard library sync package.
This will make it easier to swap out synchronization primitives. For example,
this will allow us to use primitives from github.com/sasha-s/go-deadlock to
check for lock ordering violations.
Updates #1472
PiperOrigin-RevId: 289033387
|
|
- Make FilesystemImpl methods that operate on parent directories require
!rp.Done() (i.e. there is at least one path component to resolve) as
precondition and postcondition (in cases where they do not finish path
resolution due to mount boundary / absolute symlink), and require that they
do not need to follow the last path component (the file being created /
deleted) as a symlink. Check for these in VFS.
- Add FilesystemImpl.GetParentDentryAt(), which is required to obtain the old
parent directory for VFS.RenameAt(). (Passing the Dentry to be renamed
instead has the wrong semantics if the file named by the old path is a mount
point since the Dentry will be on the wrong Mount.)
- Update memfs to implement these methods correctly (?), including RenameAt.
- Change fspath.Parse() to allow empty paths (to simplify implementation of
AT_EMPTY_PATH).
- Change vfs.PathOperation to take a fspath.Path instead of a raw pathname;
non-test callers will need to fspath.Parse() pathnames themselves anyway in
order to detect absolute paths and select PathOperation.Start accordingly.
PiperOrigin-RevId: 286934941
|
|
PiperOrigin-RevId: 286660774
|
|
- Remove the Filesystem argument from DentryImpl.*Ref(); in general DentryImpls
that need the Filesystem for reference counting will probably also need it
for other interface methods that don't plumb Filesystem, so it's easier to
just store a pointer to the filesystem in the DentryImpl.
- Add a pointer to the VirtualFilesystem to Filesystem, which is needed by the
gofer client to disown dentries for cache eviction triggered by dentry
reference count changes.
- Rename FilesystemType.NewFilesystem to GetFilesystem; in some cases (e.g.
sysfs, cgroupfs) it's much cleaner for there to be only one Filesystem that
is used by all mounts, and in at least one case (devtmpfs) it's visibly
incorrect not to do so, so NewFilesystem doesn't always actually create and
return a *new* Filesystem.
- Require callers of FileDescription.Init() to increment Mount/Dentry
references. This is because the gofer client may, in the OpenAt() path, take
a reference on a dentry with 0 references, which is safe due to
synchronization that is outside the scope of this CL, and it would be safer
to still have its implementation of DentryImpl.IncRef() check for an
increment for 0 references in other cases.
- Add FileDescription.TryIncRef. This is used by the gofer client to take
references on "special file descriptions" (FDs for files such as pipes,
sockets, and devices), which use per-FD handles (fids) instead of
dentry-shared handles, for sync() and syncfs().
PiperOrigin-RevId: 282473364
|
|
This is required to test filesystems with a non-trivial implementation of
FilesystemImpl.Release(). Propagation isn't handled yet, and umount isn't yet
plumbed out to VirtualFilesystem.UmountAt(), but otherwise the implementation
of umount is believed to be correct.
- Move entering mountTable.seq writer critical sections to callers of
mountTable.{insert,remove}Seqed. This is required since umount(2) must ensure
that no new references are taken on the candidate mount after checking that
it isn't busy, which is only possible by entering a vfs.mountTable.seq writer
critical section before the check and remaining in it until after
VFS.umountRecursiveLocked() is complete. (Linux does the same thing:
fs/namespace.c:do_umount() => lock_mount_hash(),
fs/pnode.c:propagate_mount_busy(), umount_tree(), unlock_mount_hash().)
- It's not possible for dentry deletion to umount while only holding
VFS.mountMu for reading, but it's also very unappealing to hold VFS.mountMu
exclusively around e.g. gofer unlink RPCs. Introduce dentry.mu to avoid these
problems. This means that VFS.mountMu is never acquired for reading, so
change it to a sync.Mutex.
PiperOrigin-RevId: 282444343
|
|
Major differences from the current ("v1") sentry VFS:
- Path resolution is Filesystem-driven (FilesystemImpl methods call
vfs.ResolvingPath methods) rather than VFS-driven (fs package owns a
Dirent tree and calls fs.InodeOperations methods to populate it). This
drastically improves performance, primarily by reducing overhead from
inefficient synchronization and indirection. It also makes it possible
to implement remote filesystem protocols that translate FS system calls
into single RPCs, rather than having to make (at least) one RPC per path
component, significantly reducing the latency of remote filesystems
(especially during cold starts and for uncacheable shared filesystems).
- Mounts are correctly represented as a separate check based on
contextual state (current mount) rather than direct replacement in a
fs.Dirent tree. This makes it possible to support (non-recursive) bind
mounts and mount namespaces.
Included in this CL is fsimpl/memfs, an incomplete in-memory filesystem
that exists primarily to demonstrate intended filesystem implementation
patterns and for benchmarking:
BenchmarkVFS1TmpfsStat/1-6 3000000 497 ns/op
BenchmarkVFS1TmpfsStat/2-6 2000000 676 ns/op
BenchmarkVFS1TmpfsStat/3-6 2000000 904 ns/op
BenchmarkVFS1TmpfsStat/8-6 1000000 1944 ns/op
BenchmarkVFS1TmpfsStat/64-6 100000 14067 ns/op
BenchmarkVFS1TmpfsStat/100-6 50000 21700 ns/op
BenchmarkVFS2MemfsStat/1-6 10000000 197 ns/op
BenchmarkVFS2MemfsStat/2-6 5000000 233 ns/op
BenchmarkVFS2MemfsStat/3-6 5000000 268 ns/op
BenchmarkVFS2MemfsStat/8-6 3000000 477 ns/op
BenchmarkVFS2MemfsStat/64-6 500000 2592 ns/op
BenchmarkVFS2MemfsStat/100-6 300000 4045 ns/op
BenchmarkVFS1TmpfsMountStat/1-6 2000000 679 ns/op
BenchmarkVFS1TmpfsMountStat/2-6 2000000 912 ns/op
BenchmarkVFS1TmpfsMountStat/3-6 1000000 1113 ns/op
BenchmarkVFS1TmpfsMountStat/8-6 1000000 2118 ns/op
BenchmarkVFS1TmpfsMountStat/64-6 100000 14251 ns/op
BenchmarkVFS1TmpfsMountStat/100-6 100000 22397 ns/op
BenchmarkVFS2MemfsMountStat/1-6 5000000 317 ns/op
BenchmarkVFS2MemfsMountStat/2-6 5000000 361 ns/op
BenchmarkVFS2MemfsMountStat/3-6 5000000 387 ns/op
BenchmarkVFS2MemfsMountStat/8-6 3000000 582 ns/op
BenchmarkVFS2MemfsMountStat/64-6 500000 2699 ns/op
BenchmarkVFS2MemfsMountStat/100-6 300000 4133 ns/op
From this we can infer that, on this machine:
- Constant cost for tmpfs stat() is ~160ns in VFS2 and ~280ns in VFS1.
- Per-path-component cost is ~35ns in VFS2 and ~215ns in VFS1, a
difference of about 6x.
- The cost of crossing a mount boundary is about 80ns in VFS2
(MemfsMountStat/1 does approximately the same amount of work as
MemfsStat/2, except that it also crosses a mount boundary). This is an
inescapable cost of the separate mount lookup needed to support bind
mounts and mount namespaces.
PiperOrigin-RevId: 258853946
|
