| Age | Commit message (Collapse) | Author |
|---|
|
Linux 4.18 and later make reads and writes coherent between pre-copy-up and
post-copy-up FDs representing the same file on an overlay filesystem. However,
memory mappings remain incoherent:
- Documentation/filesystems/overlayfs.rst, "Non-standard behavior": "If a file
residing on a lower layer is opened for read-only and then memory mapped with
MAP_SHARED, then subsequent changes to the file are not reflected in the
memory mapping."
- fs/overlay/file.c:ovl_mmap() passes through to the underlying FD without any
management of coherence in the overlay.
- Experimentally on Linux 5.2:
```
$ cat mmap_cat_page.c
#include <err.h>
#include <fcntl.h>
#include <stdio.h>
#include <string.h>
#include <sys/mman.h>
#include <unistd.h>
int main(int argc, char **argv) {
if (argc < 2) {
errx(1, "syntax: %s [FILE]", argv[0]);
}
const int fd = open(argv[1], O_RDONLY);
if (fd < 0) {
err(1, "open(%s)", argv[1]);
}
const size_t page_size = sysconf(_SC_PAGE_SIZE);
void* page = mmap(NULL, page_size, PROT_READ, MAP_SHARED, fd, 0);
if (page == MAP_FAILED) {
err(1, "mmap");
}
for (;;) {
write(1, page, strnlen(page, page_size));
if (getc(stdin) == EOF) {
break;
}
}
return 0;
}
$ gcc -O2 -o mmap_cat_page mmap_cat_page.c
$ mkdir lowerdir upperdir workdir overlaydir
$ echo old > lowerdir/file
$ sudo mount -t overlay -o "lowerdir=lowerdir,upperdir=upperdir,workdir=workdir" none overlaydir
$ ./mmap_cat_page overlaydir/file
old
^Z
[1]+ Stopped ./mmap_cat_page overlaydir/file
$ echo new > overlaydir/file
$ cat overlaydir/file
new
$ fg
./mmap_cat_page overlaydir/file
old
```
Therefore, while the VFS1 gofer client's behavior of reopening read FDs is only
necessary pre-4.18, replacing existing memory mappings (in both sentry and
application address spaces) with mappings of the new FD is required regardless
of kernel version, and this latter behavior is common to both VFS1 and VFS2.
Re-document accordingly, and change the runsc flag to enabled by default.
New test:
- Before this CL: https://source.cloud.google.com/results/invocations/5b222d2c-e918-4bae-afc4-407f5bac509b
- After this CL: https://source.cloud.google.com/results/invocations/f28c747e-d89c-4d8c-a461-602b33e71aab
PiperOrigin-RevId: 311361267
|
|
Fixes #2651.
PiperOrigin-RevId: 311193661
|
|
Synthetic sockets do not have the race condition issue in VFS2, and we will
get rid of privateunixsocket as well.
Fixes #1200.
PiperOrigin-RevId: 310386474
|
|
PiperOrigin-RevId: 310179277
|
|
p9.NoUID/GID (== uint32(-1) == auth.NoID) is not a valid auth.KUID/KGID; in
particular, using it for file ownership causes capabilities to be ineffective
since file capabilities require that the file's KUID and KGID are mapped into
the capability holder's user namespace [1], and auth.NoID is not mapped into
any user namespace. Map p9.NoUID/GID to a different, valid KUID/KGID; in the
unlikely case that an application actually using the overflow KUID/KGID
attempts an operation that is consequently permitted by client permission
checks, the remote operation will still fail with EPERM.
Since this changes the VFS2 gofer client to no longer ignore the invalid IDs
entirely, this CL both permits and requires that we change synthetic mount point
creation to use root credentials.
[1] See fs.Inode.CheckCapability or vfs.GenericCheckPermissions.
PiperOrigin-RevId: 309856455
|
|
Named pipes and sockets can be represented in two ways in gofer fs:
1. As a file on the remote filesystem. In this case, all file operations are
passed through 9p.
2. As a synthetic file that is internal to the sandbox. In this case, the
dentry stores an endpoint or VFSPipe for sockets and pipes respectively,
which replaces interactions with the remote fs through the gofer.
In gofer.filesystem.MknodAt, we attempt to call mknod(2) through 9p,
and if it fails, fall back to the synthetic version.
Updates #1200.
PiperOrigin-RevId: 308828161
|
|
The FileDescription implementation for hostfs sockets uses the standard Unix
socket implementation (unix.SocketVFS2), but is also tied to a hostfs dentry.
Updates #1672, #1476
PiperOrigin-RevId: 308716426
|
|
Fixes #1477.
PiperOrigin-RevId: 308317511
|
|
PiperOrigin-RevId: 308164359
|
|
PiperOrigin-RevId: 307977689
|
|
- Fix defer operation ordering in kernfs.Filesystem.AccessAt()
- Add AT_NULL entry in proc/pid/auvx
- Fix line padding in /proc/pid/maps
- Fix linux_dirent serialization for getdents(2)
- Remove file creation flags from vfs.FileDescription.statusFlags()
Updates #1193, #1035
PiperOrigin-RevId: 307704159
|
|
Sentry metrics with nanoseconds units are labeled as such, and non-cumulative
sentry metrics are supported.
PiperOrigin-RevId: 307621080
|
|
The comments in the ticket indicate that this behavior
is fine and that the ticket should be closed, so we shouldn't
need pointers to the ticket.
PiperOrigin-RevId: 306266071
|
|
The sentry doesn't allow execve, but it's a good defense
in-depth measure.
PiperOrigin-RevId: 305958737
|
|
PiperOrigin-RevId: 305807868
|
|
PiperOrigin-RevId: 305794509
|
|
PiperOrigin-RevId: 305592245
|
|
PiperOrigin-RevId: 305588941
|
|
Determine system time from within the sentry rather than relying on the remote
filesystem to prevent inconsistencies.
Resolve related TODOs; the time discrepancies in question don't exist anymore.
PiperOrigin-RevId: 305557099
|
|
Updates #164
PiperOrigin-RevId: 305544029
|
|
Updates #2243
|
|
This required minor restructuring of how system call tables were saved
and restored, but it makes way more sense this way.
Updates #2243
|
|
PiperOrigin-RevId: 304508083
|
|
When copyUp fails, we attempt to clean up the upper filesystem by removing any
files that have already been copied-up. If the cleanup fails, we panic because
the "overlay filesystem is in an inconsistent state".
This CL adds the original copy-up error to the panic information, to hopefully
make it easier to track down how the overlay filesystem got into the
inconsistent state.
PiperOrigin-RevId: 304053370
|
|
Using the host-defined file owner matches VFS1. It is more correct to use the
host-defined mode, since the cached value may become out of date. However,
kernfs.Inode.Mode() does not return an error--other filesystems on kernfs are
in-memory so retrieving mode should not fail. Therefore, if the host syscall
fails, we rely on a cached value instead.
Updates #1672.
PiperOrigin-RevId: 303220864
|
|
Frozen was to lock down changes to the host filesystem
for hostFS. Now that hostFS is gone, it can be removed.
PiperOrigin-RevId: 301907923
|
|
Refactor fs/host.TTYFileOperations so that the relevant functionality can be
shared with VFS2 (fsimpl/host.ttyFD).
Incorporate host.defaultFileFD into the default host.fileDescription. This way,
there is no need for a separate default_file.go. As in vfs1, the TTY file
implementation can be built on top of this default and override operations as
necessary (PRead/Read/PWrite/Write, Release, Ioctl).
Note that these changes still need to be plumbed into runsc, which refers to
imported TTYs in control/proc.go:ExecAsync.
Updates #1672.
PiperOrigin-RevId: 301718157
|
|
PiperOrigin-RevId: 301402181
|
|
- When setting up the virtual filesystem, mount a host.filesystem to contain
all files that need to be imported.
- Make read/preadv syscalls to the host in cases where preadv2 may not be
supported yet (likewise for writing).
- Make save/restore functions in kernel/kernel.go return early if vfs2 is
enabled.
PiperOrigin-RevId: 300922353
|
|
- Make oomScoreAdj a ThreadGroup field (Linux: signal_struct::oom_score_adj).
- Avoid deadlock caused by Task.OOMScoreAdj()/SetOOMScoreAdj() locking Task.mu
and TaskSet.mu in the wrong order (via Task.ExitState()).
PiperOrigin-RevId: 300814698
|
|
Issue #1833
PiperOrigin-RevId: 299998105
|
|
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
|
|
When list elements are removed from a list but not discarded, it becomes
important to invalidate the references they hold to their former
neighbors to prevent memory leaks.
PiperOrigin-RevId: 299412421
|
|
Adds an oom_score_adj and oom_score proc file stub. oom_score_adj accepts
writes of values -1000 to 1000 and persists the value with the task. New tasks
inherit the parent's oom_score_adj.
oom_score is a read-only stub that always returns the value '0'.
Issue #202
PiperOrigin-RevId: 299245355
|
|
PiperOrigin-RevId: 298380654
|
|
/dev/net/tun does not currently work with hostinet. This has caused some
program starts failing because it thinks the feature exists.
PiperOrigin-RevId: 297876196
|
|
MayDelete must lock the directory also, otherwise concurrent renames may
race. Note that this also changes the methods to be aligned with the actual
Remove and RemoveDirectory methods to minimize confusion when reading the
code. (It was hard to see that resolution was correct.)
PiperOrigin-RevId: 297258304
|
|
Each mount is holds a reference on a root Dirent, but the mount itself may
live beyond it's own reference. This means that a call to Root() can come
after the associated reference has been dropped.
Instead of introducing a separate layer of references for mount objects,
we simply change the Root() method to use TryIncRef() and allow it to return
nil if the mount is already gone. This requires updating a small number of
callers and minimizes the change (since VFSv2 will replace this code shortly).
PiperOrigin-RevId: 297174230
|
|
PiperOrigin-RevId: 296526279
|
|
TCP/IP will work with netstack networking. hostinet doesn't work, and sockets
will have the same behavior as it is now.
Before the userspace is able to create device, the default loopback device can
be used to test.
/proc/net and /sys/net will still be connected to the root network stack; this
is the same behavior now.
Issue #1833
PiperOrigin-RevId: 296309389
|
|
- 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
|
|
Fixes #1812. (The more direct cause of the deadlock is panic unsafety because
the historically high cost of defer means that we avoid it in hot paths,
including much of MM; defer is much cheaper as of Go 1.14, but still a
measurable overhead.)
PiperOrigin-RevId: 294560316
|
|
The slaveInodeOperations is currently copying the object when
truncate is called (which is a no-op). This may result in a
(unconsequential) data race when being modified concurrently.
PiperOrigin-RevId: 294484276
|
|
PiperOrigin-RevId: 294295852
|
|
Note that these are only implemented for tmpfs, and other impls will still
return EOPNOTSUPP.
PiperOrigin-RevId: 293899385
|
|
These were out-of-band notes that can help provide additional context
and simplify automated imports.
PiperOrigin-RevId: 293525915
|
|
Updates #1198
Opening host pipes (by spinning in fdpipe) and host sockets is not yet
complete, and will be done in a future CL.
Major differences from VFS1 gofer client (sentry/fs/gofer), with varying levels
of backportability:
- "Cache policies" are replaced by InteropMode, which control the behavior of
timestamps in addition to caching. Under InteropModeExclusive (analogous to
cacheAll) and InteropModeWritethrough (analogous to cacheAllWritethrough),
client timestamps are *not* written back to the server (it is not possible in
9P or Linux for clients to set ctime, so writing back client-authoritative
timestamps results in incoherence between atime/mtime and ctime). Under
InteropModeShared (analogous to cacheRemoteRevalidating), client timestamps
are not used at all (remote filesystem clocks are authoritative). cacheNone
is translated to InteropModeShared + new option
filesystemOptions.specialRegularFiles.
- Under InteropModeShared, "unstable attribute" reloading for permission
checks, lookup, and revalidation are fused, which is feasible in VFS2 since
gofer.filesystem controls path resolution. This results in a ~33% reduction
in RPCs for filesystem operations compared to cacheRemoteRevalidating. For
example, consider stat("/foo/bar/baz") where "/foo/bar/baz" fails
revalidation, resulting in the instantiation of a new dentry:
VFS1 RPCs:
getattr("/") // fs.MountNamespace.FindLink() => fs.Inode.CheckPermission() => gofer.inodeOperations.check() => gofer.inodeOperations.UnstableAttr()
walkgetattr("/", "foo") = fid1 // fs.Dirent.walk() => gofer.session.Revalidate() => gofer.cachePolicy.Revalidate()
clunk(fid1)
getattr("/foo") // CheckPermission
walkgetattr("/foo", "bar") = fid2 // Revalidate
clunk(fid2)
getattr("/foo/bar") // CheckPermission
walkgetattr("/foo/bar", "baz") = fid3 // Revalidate
clunk(fid3)
walkgetattr("/foo/bar", "baz") = fid4 // fs.Dirent.walk() => gofer.inodeOperations.Lookup
getattr("/foo/bar/baz") // linux.stat() => gofer.inodeOperations.UnstableAttr()
VFS2 RPCs:
getattr("/") // gofer.filesystem.walkExistingLocked()
walkgetattr("/", "foo") = fid1 // gofer.filesystem.stepExistingLocked()
clunk(fid1)
// No getattr: walkgetattr already updated metadata for permission check
walkgetattr("/foo", "bar") = fid2
clunk(fid2)
walkgetattr("/foo/bar", "baz") = fid3
// No clunk: fid3 used for new gofer.dentry
// No getattr: walkgetattr already updated metadata for stat()
- gofer.filesystem.unlinkAt() does not require instantiation of a dentry that
represents the file to be deleted. Updates #898.
- gofer.regularFileFD.OnClose() skips Tflushf for regular files under
InteropModeExclusive, as it's nonsensical to request a remote file flush
without flushing locally-buffered writes to that remote file first.
- Symlink targets are cached when InteropModeShared is not in effect.
- p9.QID.Path (which is already required to be unique for each file within a
server, and is accordingly already synthesized from device/inode numbers in
all known gofers) is used as-is for inode numbers, rather than being mapped
along with attr.RDev in the client to yet another synthetic inode number.
- Relevant parts of fsutil.CachingInodeOperations are inlined directly into
gofer package code. This avoids having to duplicate part of its functionality
in fsutil.HostMappable.
PiperOrigin-RevId: 293190213
|
|
Internal pipes are supported similarly to how internal UDS is done.
It is also controlled by the same flag.
Fixes #1102
PiperOrigin-RevId: 293150045
|
|
PiperOrigin-RevId: 292587459
|
|
Splice must not allow negative offsets. Writes also must not allow offset +
size to overflow int64. Reads are similarly broken, but not just in splice
(b/148095030).
Reported-by: syzbot+0e1ff0b95fb2859b4190@syzkaller.appspotmail.com
PiperOrigin-RevId: 292361208
|
