hyperguild/ingestion/internal/brainfs/brainfs.go

// Package brainfs reads the brain directory through an os.Root (hyperguild#157).
// The brain dir is a Syncthing sendreceive folder and Syncthing syncs
// symlinks; a reader that walks and reads by path follows a planted symlink
// out of the brain, and blocks forever on a planted FIFO. Readers that go
// through a Brain only ever read regular files inside the brain: os.Root
// refuses to leave the root even if an entry is swapped mid-read, and every
// open is non-blocking with the type checked on the opened handle.
package brainfs

import (
	"errors"
	"fmt"
	"io"
	"io/fs"
	"os"
	"path"
	"path/filepath"
	"strings"
	"syscall"
)

// ErrNotRegular is returned for an entry that is not a regular file.
var ErrNotRegular = errors.New("not a regular file")

// ErrTooLarge is returned by ReadRegularMax for a file over the limit.
var ErrTooLarge = errors.New("file too large")

// Brain is an open brain directory.
type Brain struct {
	root *os.Root
	dir  string
}

// Open opens brainDir (which may itself be a mount symlink) as a root.
func Open(brainDir string) (*Brain, error) {
	r, err := os.OpenRoot(brainDir)
	if err != nil {
		return nil, fmt.Errorf("open brain root: %w", err)
	}
	return &Brain{root: r, dir: brainDir}, nil
}

// Close releases the root.
func (b *Brain) Close() error { return b.root.Close() }

// Rel converts a path under the brain dir to a root-relative path.
func (b *Brain) Rel(path string) (string, error) {
	rel, err := filepath.Rel(b.dir, path)
	if err != nil {
		return "", err
	}
	if rel == ".." || len(rel) > 2 && rel[:3] == ".."+string(filepath.Separator) {
		return "", fmt.Errorf("%s is outside the brain dir", path)
	}
	return filepath.ToSlash(rel), nil
}

// Lstat stats a brain-relative path through the root without following a
// final symlink.
func (b *Brain) Lstat(rel string) (fs.FileInfo, error) {
	return b.root.Lstat(filepath.ToSlash(rel))
}

// realDir returns ErrNotRegular unless every component of rel is a real
// directory. A swap after the check is not caught (accepted residual,
// hyperguild#157).
func (b *Brain) realDir(rel string) error {
	if rel == "." {
		return nil
	}
	prefix := ""
	for _, part := range strings.Split(rel, "/") {
		prefix = path.Join(prefix, part)
		fi, err := b.root.Lstat(prefix)
		if err != nil {
			return err
		}
		if !fi.IsDir() {
			return fmt.Errorf("%s: walk root component %s is not a directory: %w", rel, prefix, ErrNotRegular)
		}
	}
	return nil
}

// ReadDirRegular lists the regular files directly in a brain-relative
// directory, by name. Every component of relDir must be a real directory
// (ErrNotRegular otherwise); symlinks, FIFOs, devices and subdirectories in
// it are left out.
func (b *Brain) ReadDirRegular(relDir string) ([]string, error) {
	relDir = filepath.ToSlash(relDir)
	if err := b.realDir(relDir); err != nil {
		return nil, err
	}
	ents, err := fs.ReadDir(b.root.FS(), relDir)
	if err != nil {
		return nil, err
	}
	var names []string
	for _, e := range ents {
		if e.Type().IsRegular() {
			names = append(names, e.Name())
		}
	}
	return names, nil
}

// ReadRegular reads a brain-relative path, refusing anything that is not a
// regular file, both before the open and on the opened handle.
func (b *Brain) ReadRegular(rel string) ([]byte, error) {
	fi, err := b.root.Lstat(rel)
	if err != nil {
		return nil, err
	}
	if !fi.Mode().IsRegular() {
		return nil, fmt.Errorf("%s: %w", rel, ErrNotRegular)
	}
	return b.readOpenedMax(rel, -1, fi)
}

// readOpened opens rel non-blocking and reads it only if the OPENED handle is
// a regular file. This is the check that matters under a race: an entry
// swapped for a FIFO after the Lstat, with a writer holding it open, would
// otherwise park io.ReadAll forever (PR #161 review: removing it hung search
// after 16 iterations).
//
// Accepted residual, as in api/rawroot.go: a HARD link to an outside file is
// a regular file and is read. Syncthing syncs contents, not links, so a peer
// cannot create one; fs.protected_hardlinks=1 and the pod's secrets are on a
// different filesystem from the brain volume, so a local caller cannot link
// them in. Refusing Nlink > 1 is the next step if either changes.
func (b *Brain) readOpened(rel string) ([]byte, error) { return b.readOpenedMax(rel, -1, nil) }

// ReadRegularMax is ReadRegular with a size limit, checked on the opened
// handle before reading (max < 0 means no limit).
func (b *Brain) ReadRegularMax(rel string, max int64) ([]byte, error) {
	fi, err := b.root.Lstat(rel)
	if err != nil {
		return nil, err
	}
	if !fi.Mode().IsRegular() {
		return nil, fmt.Errorf("%s: %w", rel, ErrNotRegular)
	}
	return b.readOpenedMax(rel, max, fi)
}

// readOpenedMax opens rel and reads it if the opened handle is a regular file
// and, when lfi is set, the same file the caller Lstat-ed: os.Root follows an
// in-root final symlink even with O_NOFOLLOW, so a symlink to another brain
// file swapped in between the Lstat and the open is refused only here.
func (b *Brain) readOpenedMax(rel string, max int64, lfi fs.FileInfo) ([]byte, error) {
	f, err := b.root.OpenFile(rel, os.O_RDONLY|syscall.O_NONBLOCK, 0)
	if err != nil {
		return nil, err
	}
	defer func() { _ = f.Close() }()
	hfi, err := f.Stat()
	if err != nil {
		return nil, err
	}
	if !hfi.Mode().IsRegular() {
		return nil, fmt.Errorf("%s: %w", rel, ErrNotRegular)
	}
	if lfi != nil && !os.SameFile(lfi, hfi) {
		return nil, fmt.Errorf("%s: changed while opening: %w", rel, ErrNotRegular)
	}
	if max >= 0 && hfi.Size() > max {
		return nil, fmt.Errorf("%s is %d bytes (limit %d): %w", rel, hfi.Size(), max, ErrTooLarge)
	}
	var r io.Reader = f
	if wrapAfterStat != nil {
		r = wrapAfterStat(r)
	}
	if max < 0 {
		return io.ReadAll(r)
	}
	// The handle's size can grow after the check; bound the read itself.
	data, err := io.ReadAll(io.LimitReader(r, max+1))
	if err != nil {
		return nil, err
	}
	if int64(len(data)) > max {
		return nil, fmt.Errorf("%s grew past %d bytes while read: %w", rel, max, ErrTooLarge)
	}
	return data, nil
}

// wrapAfterStat is a test seam: it runs between the size check and the read,
// and wraps the reader the read consumes. A package global: brainfs tests
// must not use t.Parallel.
var wrapAfterStat func(io.Reader) io.Reader

// WalkRegular walks a brain-relative directory through the root and calls fn
// for each regular file. Directories are descended; symlinks, FIFOs and
// devices found by the walk are skipped. A walk root with any component that
// is not a real directory goes to onErr as ErrNotRegular and nothing is
// yielded. An entry error is passed to onErr and the walk continues; a nil
// onErr ignores it.
//
// Accepted residual (hyperguild#157): a directory swapped for a symlink
// between the listing and the read is followed. os.Root still keeps the read
// inside the brain, but not inside relDir.
func (b *Brain) WalkRegular(relDir string, fn func(rel string) error, onErr func(rel string, err error)) error {
	relDir = filepath.ToSlash(relDir)
	// fs.WalkDir stats its root with a follow, and os.Root allows a symlink
	// that stays inside the brain: knowledge -> . would alias the whole
	// brain. Require every component of the root to be a real directory
	// (Lstat checks only the last one).
	err := b.realDir(relDir)
	if err != nil {
		if onErr != nil {
			onErr(relDir, err)
		}
		return nil
	}
	return fs.WalkDir(b.root.FS(), relDir, func(p string, d fs.DirEntry, err error) error {
		if err != nil {
			if onErr != nil {
				onErr(p, err)
			}
			if d != nil && d.IsDir() {
				return fs.SkipDir
			}
			return nil
		}
		if d.IsDir() || !d.Type().IsRegular() {
			return nil
		}
		return fn(p)
	})
}