These inodes can never be part of a filesystem tree. They are nameless and
never have a parent.
This allows us to avoid taking a lock in kernfs.InotifyWithParent for such
anonymous inodes.
PiperOrigin-RevId: 538823227
This commit adds support for program execution via a host file
descriptor. To make use of this feature, the host file descriptor must
be provided to the --exec-fd argument. For example,
exec 3</usr/bin/echo
runsc exec --exec-fd=3 mycontainer hello world
will run the host's echo binary inside gVisor. In this case, "hello" is
supplied to echo as argv[0]. As a result, the output of the above
command is "world".
This feature is useful for bootstrapping unknown guest environments and
allows static executables to perform setup actions inside the container
while they need not be part of the guest file system.
This catches up the interface to the `EmitUnimplementedEvent` method signature
on `kernel.Kernel`.
Also add build-time test to verify that `kernel.Kernel` implements this
interface, in order to catch such breakages at build time in the future.
PiperOrigin-RevId: 519000411
This is consistent with Linux. Some applications recursively add
files to inotify FD. They fail if they run into kernfs files
because inotify_add_watch(2) returns EPERM in that case.
PiperOrigin-RevId: 462642204
fchmod(2) is not part of sentry's seccomp filters. So the sentry
should not be trying to make this syscall. Instead we will error
out gracefully. Violating seccomp filters will kill the entire
sandbox.
There was only one instance in host filesystem (both in VFS1
and VFS2). I am only fixing VFS2, because VFS1 is already
deprecated. I manually checked that all other host syscalls
made by fsimpl/host are allowed by sentry seccomp filters.
PiperOrigin-RevId: 460335687
Since refsvfs2 are used in netstack, we should use atomicbitops to avoid
breaking 32-bit builds.
On 64-bit builds there is no performance difference.
PiperOrigin-RevId: 439687980
Linux behaves differently for regular files and dirs for poll(2)/select(2)
compared to epoll_ctl(2). The latter returns EPERM for file and dirs.
I've also changed host FDs to behave like the underlying FD in regards
to epoll to keep it compatible with docker.
Fixes#7134
PiperOrigin-RevId: 429412692
Docker maps stdin to `/dev/null` which doesn't support epoll. Host FD
was ignoring the error and suceeding the epoll_ctl call from the
container, giving false impressing that epoll would be notified.
This required plumbing failure to all waiter.Waitable.EventRegister
callers and implementers.
Closes#6795
PiperOrigin-RevId: 414797621
This is needed so that connectioned endpoint in the transport package can use
this endpoint to implement host FD based binded endpoints.
I had to simplify some other dependencies to make this possible.
- Removed uniqueid's dependency on transport package completely.
- Removed SCMConnectedEndpoint and HostConnectedEndpoint's dependency on
control package so they could be moved to transport. control already depends
on transport.
- scmRights struct from fsimpl/host/control.go had to be moved into transport
so that HostConnectedEndpoint could be implemented. But scmRights.Fill()
could not be moved because it inherently depends on making
vfs.FileDescriptions which depends on vfs which in turn depends on transport.
So now that scmRights -> vfs.FD conversion happens in the syscall package.
PiperOrigin-RevId: 413839350
Instead of passing the event mask at registratrion time, pass the mask as part
of the waiter. This makes the mask immutable and simplifies the architecture of
waiters. This is also necessary for a future fix that will allow the fdnotifier
to keep persistent entries, as opposed to requiring constant updates.
This change is intended to be a no-op in terms of function. The only exception
is signalfd, where this mask was abused. To handle this case, the operation of
signalfd changed to allow one layer of indirection.
PiperOrigin-RevId: 409702998
Removes package syserror and moves still relevant code to either linuxerr
or to syserr (to be later removed).
Internal errors are converted from random types to *errors.Error types used
in linuxerr. Internal errors are in linuxerr/internal.go.
PiperOrigin-RevId: 390724202
Set stdio ownership based on the container's user to ensure the
user can open/read/write to/from stdios.
1. stdios in the host are changed to have the owner be the same
uid/gid of the process running the sandbox. This ensures that the
sandbox has full control over it.
2. stdios owner owner inside the sandbox is changed to match the
container's user to give access inside the container and make it
behave the same as runc.
Fixes#6180
PiperOrigin-RevId: 384347009
Update all instances of the above errors to the faster linuxerr implementation.
With the temporary linuxerr.Equals(), no logical changes are made.
PiperOrigin-RevId: 382306655
Remove three syserror entries duplicated in linuxerr. Because of the
linuxerr.Equals method, this is a mere change of return values from
syserror to linuxerr definitions.
Done with only these three errnos as CLs removing all grow to a significantly
large size.
PiperOrigin-RevId: 382173835
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
On Linux these are meant to be equivalent to POLLIN/POLLOUT. Rather
than hack these on in sys_poll etc it felt cleaner to just cleanup
the call sites to notify for both events. This is what linux does
as well.
Fixes#5544
PiperOrigin-RevId: 364859977
The syscall package has been deprecated in favor of golang.org/x/sys.
Note that syscall is still used in the following places:
- pkg/sentry/socket/hostinet/stack.go: some netlink related functionalities
are not yet available in golang.org/x/sys.
- syscall.Stat_t is still used in some places because os.FileInfo.Sys() still
returns it and not unix.Stat_t.
Updates #214
PiperOrigin-RevId: 360701387
The limits for snd/rcv buffers for unix domain socket is controlled by the
following sysctls on linux
- net.core.rmem_default
- net.core.rmem_max
- net.core.wmem_default
- net.core.wmem_max
Today in gVisor we do not expose these sysctls but we do support setting the
equivalent in netstack via stack.Options() method. But AF_UNIX sockets in gVisor
can be used without netstack, with hostinet or even without any networking stack
at all. Which means ideally these sysctls need to live as globals in gVisor.
But rather than make this a big change for now we hardcode the limits in the
AF_UNIX implementation itself (which in itself is better than where we were
before) where it SO_SNDBUF was hardcoded to 16KiB. Further we bump the initial
limit to a default value of 208 KiB to match linux from the paltry 16 KiB we use
today.
Updates #5132
PiperOrigin-RevId: 356665498
This lets us avoid treating a value of 0 as one reference. All references
using the refsvfs2 template must call InitRefs() before the reference is
incremented/decremented, or else a panic will occur. Therefore, it should be
pretty easy to identify missing InitRef calls during testing.
Updates #1486.
PiperOrigin-RevId: 341411151
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
Singleton filesystem like devpts and devtmpfs have a single filesystem shared
among all mounts, so they acquire a "self-reference" when initialized that
must be released when the entire virtual filesystem is released at sandbox
exit.
PiperOrigin-RevId: 336828852
This change aims to fix the memory leak issue reported in #3933.
Background:
VFS2 kernfs kept accumulating invalid dentries if those dentries were not
walked on. After substantial consideration of the problem by our team, we
decided to have an LRU cache solution. This change is the first part to that
solution, where we don't cache anything. The LRU cache can be added on top of
this.
What has changed:
- Introduced the concept of an inode tree in kernfs.OrderedChildren.
This is helpful is cases where the lifecycle of an inode is different from
that of a dentry.
- OrderedChildren now deals with initialized inodes instead of initialized
dentries. It now implements Lookup() where it constructs a new dentry
using the inode.
- OrderedChildren holds a ref on all its children inodes. With this change,
now an inode can "outlive" a dentry pointing to it. See comments in
kernfs.OrderedChildren.
- The kernfs dentry tree is solely maintained by kernfs only. Inode
implementations can not modify the dentry tree.
- Dentries that reach ref count 0 are removed from the dentry tree.
- revalidateChildLocked now defer-DecRefs the newly created dentry from
Inode.Lookup(), limiting its life to the current filesystem operation. If
refs are picked on the dentry during the FS op (via an FD or something),
then it will stick around and will be removed when the FD is closed. So there
is essentially _no caching_ for Look()ed up dentries.
- kernfs.DecRef does not have the precondition that fs.mu must be locked.
Fixes#3933
PiperOrigin-RevId: 336768576
