Add Equals method to compare syserror and unix.Errno errors to linuxerr errors.
This will facilitate removal of syserror definitions in a followup, and
finding needed conversions from unix.Errno to linuxerr.
PiperOrigin-RevId: 380909667
Split usermem package to help remove syserror dependency in go_marshal.
New hostarch package contains code not dependent on syserror.
PiperOrigin-RevId: 365651233
Inode number consistency checks are now skipped in save/restore tests for
reasons described in greatest detail in StatTest.StateDoesntChangeAfterRename.
They pass in VFS1 due to the bug described in new test case
SimpleStatTest.DifferentFilesHaveDifferentDeviceInodeNumberPairs.
Fixes#1663
PiperOrigin-RevId: 338776148
Our current reference leak checker uses finalizers to verify whether an object
has reached zero references before it is garbage collected. There are multiple
problems with this mechanism, so a rewrite is in order.
With finalizers, there is no way to guarantee that a finalizer will run before
the program exits. When an unreachable object with a finalizer is garbage
collected, its finalizer will be added to a queue and run asynchronously. The
best we can do is run garbage collection upon sandbox exit to make sure that
all finalizers are enqueued.
Furthermore, if there is a chain of finalized objects, e.g. A points to B
points to C, garbage collection needs to run multiple times before all of the
finalizers are enqueued. The first GC run will register the finalizer for A but
not free it. It takes another GC run to free A, at which point B's finalizer
can be registered. As a result, we need to run GC as many times as the length
of the longest such chain to have a somewhat reliable leak checker.
Finally, a cyclical chain of structs pointing to one another will never be
garbage collected if a finalizer is set. This is a well-known issue with Go
finalizers (https://github.com/golang/go/issues/7358). Using leak checking on
filesystem objects that produce cycles will not work and even result in memory
leaks.
The new leak checker stores reference counted objects in a global map when
leak check is enabled and removes them once they are destroyed. At sandbox
exit, any remaining objects in the map are considered as leaked. This provides
a deterministic way of detecting leaks without relying on the complexities of
finalizers and garbage collection.
This approach has several benefits over the former, including:
- Always detects leaks of objects that should be destroyed very close to
sandbox exit. The old checker very rarely detected these leaks, because it
relied on garbage collection to be run in a short window of time.
- Panics if we forgot to enable leak check on a ref-counted object (we will try
to remove it from the map when it is destroyed, but it will never have been
added).
- Can store extra logging information in the map values without adding to the
size of the ref count struct itself. With the size of just an int64, the ref
count object remains compact, meaning frequent operations like IncRef/DecRef
are more cache-efficient.
- Can aggregate leak results in a single report after the sandbox exits.
Instead of having warnings littered in the log, which were
non-deterministically triggered by garbage collection, we can print all
warning messages at once. Note that this could also be a limitation--the
sandbox must exit properly for leaks to be detected.
Some basic benchmarking indicates that this change does not significantly
affect performance when leak checking is enabled, which is understandable
since registering/unregistering is only done once for each filesystem object.
Updates #1486.
PiperOrigin-RevId: 338685972
This uses the refs_vfs2 template in vfs2 as well as objects common to vfs1 and
vfs2. Note that vfs1-only refcounts are not replaced, since vfs1 will be deleted
soon anyway.
The following structs now use the new tool, with leak check enabled:
devpts:rootInode
fuse:inode
kernfs:Dentry
kernfs:dir
kernfs:readonlyDir
kernfs:StaticDirectory
proc:fdDirInode
proc:fdInfoDirInode
proc:subtasksInode
proc:taskInode
proc:tasksInode
vfs:FileDescription
vfs:MountNamespace
vfs:Filesystem
sys:dir
kernel:FSContext
kernel:ProcessGroup
kernel:Session
shm:Shm
mm:aioMappable
mm:SpecialMappable
transport:queue
And the following use the template, but because they currently are not leak
checked, a TODO is left instead of enabling leak check in this patch:
kernel:FDTable
tun:tunEndpoint
Updates #1486.
PiperOrigin-RevId: 328460377
- Change FileDescriptionImpl Lock/UnlockPOSIX signature to
take {start,length,whence}, so the correct offset can be
calculated in the implementations.
- Create PosixLocker interface to make it possible to share
the same locking code from different implementations.
Closes#1480
PiperOrigin-RevId: 316910286
LockFD is the generic implementation that can be embedded in
FileDescriptionImpl implementations. Unique lock ID is
maintained in vfs.FileDescription and is created on demand.
Updates #1480
PiperOrigin-RevId: 315604825
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
BoundEndpointAt() is needed to support Unix sockets bound at a
file path, corresponding to BoundEndpoint() in VFS1.
Updates #1476.
PiperOrigin-RevId: 303258251
In VFS2, imported file descriptors are stored in a kernfs-based filesystem.
Upon calling ImportFD, the host fd can be accessed in two ways:
1. a FileDescription that can be added to the FDTable, and
2. a Dentry in the host.filesystem mount, which we will want to access through
magic symlinks in /proc/[pid]/fd/.
An implementation of the kernfs.Inode interface stores a unique host fd. This
inode can be inserted into file descriptions as well as dentries.
This change also plumbs in three FileDescriptionImpls corresponding to fds for
sockets, TTYs, and other files (only the latter is implemented here).
These implementations will mostly make corresponding syscalls to the host.
Where possible, the logic is ported over from pkg/sentry/fs/host.
Updates #1672
PiperOrigin-RevId: 299417263
pipe and pipe2 aren't ported, pending a slight rework of pipe FDs for VFS2.
mount and umount2 aren't ported out of temporary laziness. access and faccessat
need additional FSImpl methods to implement properly, but are stubbed to
prevent googletest from CHECK-failing. Other syscalls require additional
plumbing.
Updates #1623
PiperOrigin-RevId: 297188448
- Added fsbridge package with interface that can be used to open
and read from VFS1 and VFS2 files.
- Converted ELF loader to use fsbridge
- Added VFS2 types to FSContext
- Added vfs.MountNamespace to ThreadGroup
Updates #1623
PiperOrigin-RevId: 295183950
Because the abi will depend on the core types for marshalling (usermem,
context, safemem, safecopy), these need to be flattened from the sentry
directory. These packages contain no sentry-specific details.
PiperOrigin-RevId: 291811289
* 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
- Add FileDescriptionOptions.UseDentryMetadata, which reduces the amount of
boilerplate needed for device FDs and the like between filesystems.
- Switch back to having FileDescription.Init() take references on the Mount and
Dentry; otherwise managing refcounts around failed calls to
OpenDeviceSpecialFile() / Device.Open() is tricky.
PiperOrigin-RevId: 287575574
The former is needed for vfs.FileDescription to implement
memmap.MappingIdentity, and the latter is needed to implement getcwd(2).
PiperOrigin-RevId: 285051855
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
Zach Koopmans
gVisor bot
Jamie Liu
Dean Deng
Rahat Mahmood
Fabricio Voznika
Nicolas Lacasse
Adin Scannell
Ian Gudger
Michael Pratt