Files
Jordan MontgomeryandCopilot Autofix powered by AI c686f574fe Make the plist parser slightly more tolerant (#47371)
<!-- Add the related story/sub-task/bug number, like Resolves #123, or
remove if NA -->
**Related issue:** Resolves #

# Checklist for submitter

If some of the following don't apply, delete the relevant line.

No changes file, already covered

- [x] Input data is properly validated, `SELECT *` is avoided, SQL
injection is prevented (using placeholders for values in statements), JS
inline code is prevented especially for url redirects, and untrusted
data interpolated into shell scripts/commands is validated against shell
metacharacters.
- [x] Timeouts are implemented and retries are limited to avoid infinite
loops
- [x] If paths of existing endpoints are modified without backwards
compatibility, checked the frontend/CLI for any necessary changes

## Testing

- [x] Added/updated automated tests
- [x] Where appropriate, [automated tests simulate multiple hosts and
test for host
isolation](https://github.com/fleetdm/fleet/blob/main/docs/Contributing/reference/patterns-backend.md#unit-testing)
(updates to one hosts's records do not affect another)

- [x] QA'd all new/changed functionality manually


<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit

* **Bug Fixes**
* Improved compatibility with multiple Apple plist format versions for
more reliable file handling.
* Strengthened validation of plist inputs to reduce false positives and
improve security.
* Adjusted early bounds checking so very short or malformed plist files
are handled more robustly.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->

---------

Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
2026-06-10 17:31:46 -04:00

260 lines
7.7 KiB
Go

package apple_mdm
import (
"bytes"
"encoding/binary"
"errors"
"fmt"
"github.com/micromdm/plist"
)
// Limits applied to binary property lists before decoding. They are generous
// for the device-info plists these endpoints handle, which are a single flat
// dictionary of scalar values.
const (
// binaryPlistMagic is the prefix that selects the binary plist decoder.
// Apple doesn't fully document all versions, but "bplist00" and "bplist01" are known.
// Using "bplist0" lets us accept any "bplist0<digit>" version header.
binaryPlistMagic = "bplist0"
plistTrailerSize = 32
maxPlistObjects = 1 << 16 // distinct objects (offset-table size)
maxPlistDepth = 16 // reference nesting
maxPlistNodes = 1 << 16 // objects after references are expanded
)
var (
errPlistTooComplex = errors.New("plist exceeds parsing limits")
errMalformedPlist = errors.New("malformed binary plist")
)
// BoundedPlistUnmarshal decodes a plist into v. Binary plists are first checked
// against the depth, object-count, and object-size limits above; XML plists are
// decoded directly (their size is bounded by the caller's body limit).
func BoundedPlistUnmarshal(data []byte, v any) error {
if bytes.HasPrefix(data, []byte(binaryPlistMagic)) {
if err := checkBinaryPlistBounds(data); err != nil {
return err
}
}
return plist.Unmarshal(data, v)
}
// checkBinaryPlistBounds walks a binary plist's object references, rejecting
// input that exceeds the limits or points outside the data region.
func checkBinaryPlistBounds(data []byte) error {
// See comment on binaryPlistMagic above for why the +1 is needed
if len(data) < len(binaryPlistMagic)+1+plistTrailerSize {
return fmt.Errorf("%w: shorter than minimum size", errMalformedPlist)
}
// Trailer is the final 32 bytes (CFBinaryPlistTrailer).
trailer := data[len(data)-plistTrailerSize:]
offsetIntSize := trailer[6]
objectRefSize := trailer[7]
numObjects := binary.BigEndian.Uint64(trailer[8:16])
rootObject := binary.BigEndian.Uint64(trailer[16:24])
offsetTableOffset := binary.BigEndian.Uint64(trailer[24:32])
if offsetIntSize == 0 || offsetIntSize > 8 || objectRefSize == 0 || objectRefSize > 8 {
return fmt.Errorf("%w: invalid integer sizes", errMalformedPlist)
}
if numObjects == 0 {
return fmt.Errorf("%w: no objects", errMalformedPlist)
}
if numObjects > maxPlistObjects {
return fmt.Errorf("%w: %d objects", errPlistTooComplex, numObjects)
}
trailerStart := uint64(len(data) - plistTrailerSize) //nolint:gosec // dismiss G115, length is bounded above
tableBytes := numObjects * uint64(offsetIntSize)
if offsetTableOffset > trailerStart || tableBytes > trailerStart-offsetTableOffset {
return fmt.Errorf("%w: offset table out of bounds", errMalformedPlist)
}
offsetTable := make([]uint64, numObjects)
pos := offsetTableOffset
for i := range offsetTable {
offsetTable[i] = readUintBE(data[pos : pos+uint64(offsetIntSize)])
pos += uint64(offsetIntSize)
}
b := &plistBounder{
data: data,
offsetTable: offsetTable,
objectRefSize: objectRefSize,
dataEnd: offsetTableOffset, // objects live before the offset table
}
return b.visit(rootObject, 0)
}
type plistBounder struct {
data []byte
offsetTable []uint64
objectRefSize uint8
dataEnd uint64
nodes int
}
// visit walks the object at index, recursing into array and dict references.
// The depth and node guards keep the walk itself bounded: a cycle stops at
// maxPlistDepth, and reference expansion stops at maxPlistNodes.
func (b *plistBounder) visit(index uint64, depth int) error {
if depth > maxPlistDepth {
return fmt.Errorf("%w: nesting deeper than %d", errPlistTooComplex, maxPlistDepth)
}
b.nodes++
if b.nodes > maxPlistNodes {
return fmt.Errorf("%w: more than %d expanded objects", errPlistTooComplex, maxPlistNodes)
}
if index >= uint64(len(b.offsetTable)) {
return fmt.Errorf("%w: object ref %d out of range", errMalformedPlist, index)
}
offset := b.offsetTable[index]
if offset >= b.dataEnd {
return fmt.Errorf("%w: object offset out of range", errMalformedPlist)
}
// High nibble of the marker byte is the object type (CFBinaryPList.c).
cur := plistCursor{data: b.data, pos: offset, end: b.dataEnd}
marker, err := cur.readByte()
if err != nil {
return err
}
switch marker >> 4 {
case 0xa: // array: count object refs
count, err := cur.readCount(marker)
if err != nil {
return err
}
if count > cur.remaining()/uint64(b.objectRefSize) {
return fmt.Errorf("%w: array refs exceed input", errMalformedPlist)
}
return b.visitRefs(&cur, count, depth)
case 0xd: // dictionary: count key refs followed by count value refs
count, err := cur.readCount(marker)
if err != nil {
return err
}
if count > cur.remaining()/uint64(b.objectRefSize)/2 {
return fmt.Errorf("%w: dictionary refs exceed input", errMalformedPlist)
}
return b.visitRefs(&cur, 2*count, depth)
case 0x2: // real: 1<<(low nibble) bytes follow inline
nbytes := uint64(1) << (marker & 0xf)
if cur.pos+nbytes < cur.pos || cur.pos+nbytes > b.dataEnd {
return fmt.Errorf("%w: real size exceeds input", errMalformedPlist)
}
return nil
case 0x4, 0x5: // data, ASCII string: count single-byte units
return b.checkPayload(&cur, marker, 1)
case 0x6: // UTF-16 string: count two-byte units
return b.checkPayload(&cur, marker, 2)
default:
// Other types are fixed-size or non-recursive; nothing to bound.
return nil
}
}
func (b *plistBounder) visitRefs(cur *plistCursor, total uint64, depth int) error {
for range total {
ref, err := cur.readRef(b.objectRefSize)
if err != nil {
return err
}
if err := b.visit(ref, depth+1); err != nil {
return err
}
}
return nil
}
// checkPayload verifies a variable-length scalar (data or string) declares a
// payload that fits within the object region.
func (b *plistBounder) checkPayload(cur *plistCursor, marker byte, unitSize uint64) error {
count, err := cur.readCount(marker)
if err != nil {
return err
}
size := count * unitSize
if size/unitSize != count {
return fmt.Errorf("%w: object size overflow", errMalformedPlist)
}
if cur.pos+size < cur.pos || cur.pos+size > b.dataEnd {
return fmt.Errorf("%w: object size exceeds input", errMalformedPlist)
}
return nil
}
// plistCursor reads forward within a single object's bytes, bounded by end.
type plistCursor struct {
data []byte
pos uint64
end uint64
}
func (c *plistCursor) remaining() uint64 {
if c.pos >= c.end {
return 0
}
return c.end - c.pos
}
func (c *plistCursor) readByte() (byte, error) {
if c.pos >= c.end {
return 0, fmt.Errorf("%w: unexpected end of object data", errMalformedPlist)
}
v := c.data[c.pos]
c.pos++
return v, nil
}
func (c *plistCursor) readBytes(n uint64) ([]byte, error) {
if n == 0 {
return nil, nil
}
if c.pos+n < c.pos || c.pos+n > c.end {
return nil, fmt.Errorf("%w: unexpected end of object data", errMalformedPlist)
}
v := c.data[c.pos : c.pos+n]
c.pos += n
return v, nil
}
// readCount decodes the variable-length count used by data, strings, arrays, and dicts.
func (c *plistCursor) readCount(marker byte) (uint64, error) {
if marker&0xf != 0xf {
return uint64(marker & 0xf), nil
}
sizeMarker, err := c.readByte()
if err != nil {
return 0, err
}
nbytes := uint64(1) << (sizeMarker & 0x0f)
if nbytes > 8 {
return 0, fmt.Errorf("%w: invalid count size", errMalformedPlist)
}
buf, err := c.readBytes(nbytes)
if err != nil {
return 0, err
}
return readUintBE(buf), nil
}
func (c *plistCursor) readRef(size uint8) (uint64, error) {
buf, err := c.readBytes(uint64(size))
if err != nil {
return 0, err
}
return readUintBE(buf), nil
}
func readUintBE(b []byte) uint64 {
var n uint64
for _, c := range b {
n = n<<8 | uint64(c)
}
return n
}