]> git.proxmox.com Git - proxmox-backup.git/blob - src/pxar/decoder.rs
src/pxar/decoder.rs: improve docu
[proxmox-backup.git] / src / pxar / decoder.rs
1 //! *pxar* format decoder.
2 //!
3 //! This module contain the code to decode *pxar* archive files.
4
5 use failure::*;
6 use endian_trait::Endian;
7
8 use super::format_definition::*;
9
10 use std::io::{Read, Write};
11 use std::path::{Path, PathBuf};
12
13 use std::os::unix::io::AsRawFd;
14 use std::os::unix::io::RawFd;
15 use std::os::unix::io::FromRawFd;
16 use std::os::unix::ffi::{OsStringExt};
17 use std::ffi::{OsStr, OsString};
18
19 use nix::fcntl::OFlag;
20 use nix::sys::stat::Mode;
21 use nix::errno::Errno;
22 use nix::NixPath;
23
24 // This one need Read, but works without Seek
25 pub struct PxarDecoder<'a, R: Read> {
26 reader: &'a mut R,
27 skip_buffer: Vec<u8>,
28 }
29
30 const HEADER_SIZE: u64 = std::mem::size_of::<CaFormatHeader>() as u64;
31
32 impl <'a, R: Read> PxarDecoder<'a, R> {
33
34 pub fn new(reader: &'a mut R) -> Self {
35 let skip_buffer = vec![0u8; 64*1024];
36 Self { reader, skip_buffer }
37 }
38
39 fn read_item<T: Endian>(&mut self) -> Result<T, Error> {
40
41 let mut result: T = unsafe { std::mem::uninitialized() };
42
43 let buffer = unsafe { std::slice::from_raw_parts_mut(
44 &mut result as *mut T as *mut u8,
45 std::mem::size_of::<T>()
46 )};
47
48 self.reader.read_exact(buffer)?;
49
50 Ok(result.from_le())
51 }
52
53 fn read_symlink(&mut self, size: u64) -> Result<PathBuf, Error> {
54 if size < (HEADER_SIZE + 2) {
55 bail!("dectected short symlink target.");
56 }
57 let target_len = size - HEADER_SIZE;
58
59 if target_len > (libc::PATH_MAX as u64) {
60 bail!("symlink target too long ({}).", target_len);
61 }
62
63 let mut buffer = vec![0u8; target_len as usize];
64 self.reader.read_exact(&mut buffer)?;
65
66 let last_byte = buffer.pop().unwrap();
67 if last_byte != 0u8 {
68 bail!("symlink target not nul terminated.");
69 }
70
71 Ok(PathBuf::from(std::ffi::OsString::from_vec(buffer)))
72 }
73
74 fn read_filename(&mut self, size: u64) -> Result<OsString, Error> {
75 if size < (HEADER_SIZE + 2) {
76 bail!("dectected short filename");
77 }
78 let name_len = size - HEADER_SIZE;
79
80 if name_len > ((libc::FILENAME_MAX as u64) + 1) {
81 bail!("filename too long ({}).", name_len);
82 }
83
84 let mut buffer = vec![0u8; name_len as usize];
85 self.reader.read_exact(&mut buffer)?;
86
87 let last_byte = buffer.pop().unwrap();
88 if last_byte != 0u8 {
89 bail!("filename entry not nul terminated.");
90 }
91
92 if (buffer.len() == 1 && buffer[0] == b'.') || (buffer.len() == 2 && buffer[0] == b'.' && buffer[1] == b'.') {
93 bail!("found invalid filename with slashes.");
94 }
95
96 if buffer.iter().find(|b| (**b == b'/')).is_some() {
97 bail!("found invalid filename with slashes.");
98 }
99
100 let name = std::ffi::OsString::from_vec(buffer);
101 if name.is_empty() {
102 bail!("found empty filename.");
103 }
104
105 Ok(name)
106 }
107
108 fn restore_attributes(&mut self, _entry: &CaFormatEntry) -> Result<CaFormatHeader, Error> {
109
110 loop {
111 let head: CaFormatHeader = self.read_item()?;
112 match head.htype {
113 // fimxe: impl ...
114 _ => return Ok(head),
115 }
116 }
117 }
118
119 fn restore_mode(&mut self, entry: &CaFormatEntry, fd: RawFd) -> Result<(), Error> {
120
121 let mode = Mode::from_bits_truncate((entry.mode as u32) & 0o7777);
122
123 nix::sys::stat::fchmod(fd, mode)?;
124
125 Ok(())
126 }
127
128 fn restore_mode_at(&mut self, entry: &CaFormatEntry, dirfd: RawFd, filename: &OsStr) -> Result<(), Error> {
129
130 let mode = Mode::from_bits_truncate((entry.mode as u32) & 0o7777);
131
132 // NOTE: we want :FchmodatFlags::NoFollowSymlink, but fchmodat does not support that
133 // on linux (see man fchmodat). Fortunately, we can simply avoid calling this on symlinks.
134 nix::sys::stat::fchmodat(Some(dirfd), filename, mode, nix::sys::stat::FchmodatFlags::FollowSymlink)?;
135
136 Ok(())
137 }
138
139 fn restore_ugid(&mut self, entry: &CaFormatEntry, fd: RawFd) -> Result<(), Error> {
140
141 let uid = entry.uid as u32;
142 let gid = entry.gid as u32;
143
144 let res = unsafe { libc::fchown(fd, uid, gid) };
145 Errno::result(res)?;
146
147 Ok(())
148 }
149
150 fn restore_ugid_at(&mut self, entry: &CaFormatEntry, dirfd: RawFd, filename: &OsStr) -> Result<(), Error> {
151
152 let uid = entry.uid as u32;
153 let gid = entry.gid as u32;
154
155 let res = filename.with_nix_path(|cstr| unsafe {
156 libc::fchownat(dirfd, cstr.as_ptr(), uid, gid, libc::AT_SYMLINK_NOFOLLOW)
157 })?;
158 Errno::result(res)?;
159
160 Ok(())
161 }
162
163 fn restore_mtime(&mut self, entry: &CaFormatEntry, fd: RawFd) -> Result<(), Error> {
164
165 let times = nsec_to_update_timespec(entry.mtime);
166
167 let res = unsafe { libc::futimens(fd, &times[0]) };
168 Errno::result(res)?;
169
170 Ok(())
171 }
172
173 fn restore_mtime_at(&mut self, entry: &CaFormatEntry, dirfd: RawFd, filename: &OsStr) -> Result<(), Error> {
174
175 let times = nsec_to_update_timespec(entry.mtime);
176
177 let res = filename.with_nix_path(|cstr| unsafe {
178 libc::utimensat(dirfd, cstr.as_ptr(), &times[0], libc::AT_SYMLINK_NOFOLLOW)
179 })?;
180 Errno::result(res)?;
181
182 Ok(())
183 }
184
185 fn restore_device_at(&mut self, entry: &CaFormatEntry, dirfd: RawFd, filename: &OsStr, device: &CaFormatDevice) -> Result<(), Error> {
186
187 let rdev = nix::sys::stat::makedev(device.major, device.minor);
188 let mode = ((entry.mode as u32) & libc::S_IFMT) | 0o0600;
189 let res = filename.with_nix_path(|cstr| unsafe {
190 libc::mknodat(dirfd, cstr.as_ptr(), mode, rdev)
191 })?;
192 Errno::result(res)?;
193
194 Ok(())
195 }
196
197 fn restore_socket_at(&mut self, dirfd: RawFd, filename: &OsStr) -> Result<(), Error> {
198
199 let mode = libc::S_IFSOCK | 0o0600;
200 let res = filename.with_nix_path(|cstr| unsafe {
201 libc::mknodat(dirfd, cstr.as_ptr(), mode, 0)
202 })?;
203 Errno::result(res)?;
204
205 Ok(())
206 }
207
208 fn restore_fifo_at(&mut self, dirfd: RawFd, filename: &OsStr) -> Result<(), Error> {
209
210 let mode = libc::S_IFIFO | 0o0600;
211 let res = filename.with_nix_path(|cstr| unsafe {
212 libc::mkfifoat(dirfd, cstr.as_ptr(), mode)
213 })?;
214 Errno::result(res)?;
215
216 Ok(())
217 }
218
219 fn skip_bytes(&mut self, count: usize) -> Result<(), Error> {
220 let mut done = 0;
221 while done < count {
222 let todo = count - done;
223 let n = if todo > self.skip_buffer.len() { self.skip_buffer.len() } else { todo };
224 let data = &mut self.skip_buffer[..n];
225 self.reader.read_exact(data)?;
226 done += n;
227 }
228 Ok(())
229 }
230
231 /// Restore an archive into the specified directory.
232 ///
233 /// The directory is created if it does not exist.
234 pub fn restore<F>(
235 &mut self,
236 path: &Path,
237 callback: &F,
238 ) -> Result<(), Error>
239 where F: Fn(&Path) -> Result<(), Error>
240 {
241
242 let _ = std::fs::create_dir(path);
243
244 let dir = match nix::dir::Dir::open(path, nix::fcntl::OFlag::O_DIRECTORY, nix::sys::stat::Mode::empty()) {
245 Ok(dir) => dir,
246 Err(err) => bail!("unable to open target directory {:?} - {}", path, err),
247 };
248
249 self.restore_sequential(&mut path.to_owned(), &OsString::new(), &dir, callback)
250 }
251
252 fn restore_sequential<F>(
253 &mut self,
254 path: &mut PathBuf, // used for error reporting
255 filename: &OsStr, // repeats path last component
256 parent: &nix::dir::Dir,
257 callback: &F,
258 ) -> Result<(), Error>
259 where F: Fn(&Path) -> Result<(), Error>
260 {
261
262 let parent_fd = parent.as_raw_fd();
263
264 // read ENTRY first
265 let head: CaFormatHeader = self.read_item()?;
266 check_ca_header::<CaFormatEntry>(&head, CA_FORMAT_ENTRY)?;
267 let entry: CaFormatEntry = self.read_item()?;
268
269 let mode = entry.mode as u32; //fixme: upper 32bits?
270
271 let ifmt = mode & libc::S_IFMT;
272
273 if ifmt == libc::S_IFDIR {
274 let dir;
275 if filename.is_empty() {
276 dir = nix::dir::Dir::openat(parent_fd, ".", OFlag::O_DIRECTORY, Mode::empty())?;
277 } else {
278 dir = match dir_mkdirat(parent_fd, filename, true) {
279 Ok(dir) => dir,
280 Err(err) => bail!("unable to open directory {:?} - {}", path, err),
281 };
282 }
283
284 let mut head = self.restore_attributes(&entry)?;
285
286 while head.htype == CA_FORMAT_FILENAME {
287 let name = self.read_filename(head.size)?;
288 path.push(&name);
289 println!("NAME: {:?}", path);
290
291 self.restore_sequential(path, &name, &dir, callback)?;
292 path.pop();
293
294 head = self.read_item()?;
295 }
296
297 if head.htype != CA_FORMAT_GOODBYE {
298 bail!("got unknown header type inside directory entry {:016x}", head.htype);
299 }
300
301 println!("Skip Goodbye");
302 if head.size < HEADER_SIZE { bail!("detected short goodbye table"); }
303
304 self.skip_bytes((head.size - HEADER_SIZE) as usize)?;
305
306 self.restore_mode(&entry, dir.as_raw_fd())?;
307 self.restore_mtime(&entry, dir.as_raw_fd())?;
308 self.restore_ugid(&entry, dir.as_raw_fd())?;
309
310 return Ok(());
311 }
312
313 if filename.is_empty() {
314 bail!("got empty file name at {:?}", path)
315 }
316
317 if ifmt == libc::S_IFLNK {
318 // fixme: create symlink
319 //fixme: restore permission, acls, xattr, ...
320
321 let head: CaFormatHeader = self.read_item()?;
322 match head.htype {
323 CA_FORMAT_SYMLINK => {
324 let target = self.read_symlink(head.size)?;
325 println!("TARGET: {:?}", target);
326 if let Err(err) = symlinkat(&target, parent_fd, filename) {
327 bail!("create symlink {:?} failed - {}", path, err);
328 }
329 }
330 _ => {
331 bail!("got unknown header type inside symlink entry {:016x}", head.htype);
332 }
333 }
334
335 // self.restore_mode_at(&entry, parent_fd, filename)?; //not supported on symlinks
336 self.restore_ugid_at(&entry, parent_fd, filename)?;
337 self.restore_mtime_at(&entry, parent_fd, filename)?;
338
339 return Ok(());
340 }
341
342 if ifmt == libc::S_IFSOCK {
343
344 self.restore_socket_at(parent_fd, filename)?;
345
346 self.restore_mode_at(&entry, parent_fd, filename)?;
347 self.restore_ugid_at(&entry, parent_fd, filename)?;
348 self.restore_mtime_at(&entry, parent_fd, filename)?;
349
350 return Ok(());
351 }
352
353 if ifmt == libc::S_IFIFO {
354
355 self.restore_fifo_at(parent_fd, filename)?;
356
357 self.restore_mode_at(&entry, parent_fd, filename)?;
358 self.restore_ugid_at(&entry, parent_fd, filename)?;
359 self.restore_mtime_at(&entry, parent_fd, filename)?;
360
361 return Ok(());
362 }
363
364 if (ifmt == libc::S_IFBLK) || (ifmt == libc::S_IFCHR) {
365
366 let head: CaFormatHeader = self.read_item()?;
367 match head.htype {
368 CA_FORMAT_DEVICE => {
369 let device: CaFormatDevice = self.read_item()?;
370 self.restore_device_at(&entry, parent_fd, filename, &device)?;
371 }
372 _ => {
373 bail!("got unknown header type inside device entry {:016x}", head.htype);
374 }
375 }
376
377 self.restore_mode_at(&entry, parent_fd, filename)?;
378 self.restore_ugid_at(&entry, parent_fd, filename)?;
379 self.restore_mtime_at(&entry, parent_fd, filename)?;
380
381 return Ok(());
382 }
383
384 if ifmt == libc::S_IFREG {
385
386 let mut read_buffer: [u8; 64*1024] = unsafe { std::mem::uninitialized() };
387
388 let flags = OFlag::O_CREAT|OFlag::O_WRONLY|OFlag::O_EXCL;
389 let open_mode = Mode::from_bits_truncate(0o0600 | mode);
390
391 let mut file = match file_openat(parent_fd, filename, flags, open_mode) {
392 Ok(file) => file,
393 Err(err) => bail!("open file {:?} failed - {}", path, err),
394 };
395
396 let head = self.restore_attributes(&entry)?;
397
398 if head.htype != CA_FORMAT_PAYLOAD {
399 bail!("got unknown header type for file entry {:016x}", head.htype);
400 }
401
402 if head.size < HEADER_SIZE {
403 bail!("detected short payload");
404 }
405 let need = (head.size - HEADER_SIZE) as usize;
406 //self.reader.seek(SeekFrom::Current(need as i64))?;
407
408 let mut done = 0;
409 while done < need {
410 let todo = need - done;
411 let n = if todo > read_buffer.len() { read_buffer.len() } else { todo };
412 let data = &mut read_buffer[..n];
413 self.reader.read_exact(data)?;
414 file.write_all(data)?;
415 done += n;
416 }
417
418 self.restore_mode(&entry, file.as_raw_fd())?;
419 self.restore_mtime(&entry, file.as_raw_fd())?;
420 self.restore_ugid(&entry, file.as_raw_fd())?;
421
422 return Ok(());
423 }
424
425 Ok(())
426 }
427
428 /// List/Dump archive content.
429 ///
430 /// Simply print the list of contained files. This dumps archive
431 /// format details when the verbose flag is set (useful for debug).
432 pub fn dump_entry<W: std::io::Write>(
433 &mut self,
434 path: &mut PathBuf,
435 verbose: bool,
436 output: &mut W,
437 ) -> Result<(), Error> {
438
439 let print_head = |head: &CaFormatHeader| {
440 println!("Type: {:016x}", head.htype);
441 println!("Size: {}", head.size);
442 };
443
444 let head: CaFormatHeader = self.read_item()?;
445 if verbose {
446 println!("Path: {:?}", path);
447 print_head(&head);
448 } else {
449 println!("{:?}", path);
450 }
451
452 check_ca_header::<CaFormatEntry>(&head, CA_FORMAT_ENTRY)?;
453 let entry: CaFormatEntry = self.read_item()?;
454
455 if verbose {
456 println!("Mode: {:08x} {:08x}", entry.mode, (entry.mode as u32) & libc::S_IFDIR);
457 }
458 // fixme: dump attributes (ACLs, ...)
459
460 let ifmt = (entry.mode as u32) & libc::S_IFMT;
461
462 if ifmt == libc::S_IFDIR {
463
464 let mut entry_count = 0;
465
466 loop {
467 let head: CaFormatHeader = self.read_item()?;
468 if verbose {
469 print_head(&head);
470 }
471 match head.htype {
472
473 CA_FORMAT_FILENAME => {
474 let name = self.read_filename(head.size)?;
475 if verbose { println!("Name: {:?}", name); }
476 entry_count += 1;
477 path.push(&name);
478 self.dump_entry(path, verbose, output)?;
479 path.pop();
480 }
481 CA_FORMAT_GOODBYE => {
482 let table_size = (head.size - HEADER_SIZE) as usize;
483 if verbose {
484 println!("Goodbye: {:?}", path);
485 self.dump_goodby_entries(entry_count, table_size)?;
486 } else {
487 self.skip_bytes(table_size)?;
488 }
489 break;
490 }
491 _ => {
492 panic!("got unexpected header type inside directory");
493 }
494 }
495 }
496 } else {
497
498 let head: CaFormatHeader = self.read_item()?;
499 if verbose {
500 print_head(&head);
501 }
502
503 match head.htype {
504
505 CA_FORMAT_SYMLINK => {
506 let target = self.read_symlink(head.size)?;
507 if verbose {
508 println!("Symlink: {:?}", target);
509 }
510 }
511 CA_FORMAT_DEVICE => {
512 let device: CaFormatDevice = self.read_item()?;
513 if verbose {
514 println!("Device: {}, {}", device.major, device.minor);
515 }
516 }
517 CA_FORMAT_PAYLOAD => {
518 let payload_size = (head.size - HEADER_SIZE) as usize;
519 if verbose {
520 println!("Payload: {}", payload_size);
521 }
522 self.skip_bytes(payload_size)?;
523 }
524 _ => {
525 panic!("got unexpected header type inside non-directory");
526 }
527 }
528 }
529
530 Ok(())
531 }
532
533 fn dump_goodby_entries(
534 &mut self,
535 entry_count: usize,
536 table_size: usize,
537 ) -> Result<(), Error> {
538
539 let item_size = std::mem::size_of::<CaFormatGoodbyeItem>();
540 if table_size < item_size {
541 bail!("Goodbye table to small ({} < {})", table_size, item_size);
542 }
543 if (table_size % item_size) != 0 {
544 bail!("Goodbye table with strange size ({})", table_size);
545 }
546
547 let entries = table_size / item_size;
548
549 if entry_count != (entries - 1) {
550 bail!("Goodbye table with wrong entry count ({} != {})", entry_count, entries - 1);
551 }
552
553 let mut count = 0;
554
555 loop {
556 let item: CaFormatGoodbyeItem = self.read_item()?;
557 count += 1;
558 if item.hash == CA_FORMAT_GOODBYE_TAIL_MARKER {
559 if count != entries {
560 bail!("unexpected goodbye tail marker");
561 }
562 println!("Goodby tail mark.");
563 break;
564 }
565 println!("Goodby item: offset {}, size {}, hash {:016x}", item.offset, item.size, item.hash);
566 if count >= (table_size / item_size) {
567 bail!("too many goodbye items (no tail marker)");
568 }
569 }
570
571 Ok(())
572 }
573 }
574
575 fn file_openat(parent: RawFd, filename: &OsStr, flags: OFlag, mode: Mode) -> Result<std::fs::File, Error> {
576
577 let fd = filename.with_nix_path(|cstr| {
578 nix::fcntl::openat(parent, cstr.as_ref(), flags, mode)
579 })??;
580
581 let file = unsafe { std::fs::File::from_raw_fd(fd) };
582
583 Ok(file)
584 }
585
586 fn dir_mkdirat(parent: RawFd, filename: &OsStr, create_new: bool) -> Result<nix::dir::Dir, nix::Error> {
587
588 // call mkdirat first
589 let res = filename.with_nix_path(|cstr| unsafe {
590 libc::mkdirat(parent, cstr.as_ptr(), libc::S_IRWXU)
591 })?;
592
593 match Errno::result(res) {
594 Ok(_) => {},
595 Err(err) => {
596 if err == nix::Error::Sys(nix::errno::Errno::EEXIST) {
597 if create_new { return Err(err); }
598 } else {
599 return Err(err);
600 }
601 }
602 }
603
604 let dir = nix::dir::Dir::openat(parent, filename, OFlag::O_DIRECTORY, Mode::empty())?;
605
606 Ok(dir)
607 }
608
609 fn symlinkat(target: &Path, parent: RawFd, linkname: &OsStr) -> Result<(), Error> {
610
611 target.with_nix_path(|target| {
612 linkname.with_nix_path(|linkname| {
613 let res = unsafe { libc::symlinkat(target.as_ptr(), parent, linkname.as_ptr()) };
614 Errno::result(res)?;
615 Ok(())
616 })?
617 })?
618 }
619
620 fn nsec_to_update_timespec(mtime_nsec: u64) -> [libc::timespec; 2] {
621
622 // restore mtime
623 const UTIME_OMIT: i64 = ((1 << 30) - 2);
624 const NANOS_PER_SEC: i64 = 1_000_000_000;
625
626 let sec = (mtime_nsec as i64) / NANOS_PER_SEC;
627 let nsec = (mtime_nsec as i64) % NANOS_PER_SEC;
628
629 let times: [libc::timespec; 2] = [
630 libc::timespec { tv_sec: 0, tv_nsec: UTIME_OMIT },
631 libc::timespec { tv_sec: sec, tv_nsec: nsec },
632 ];
633
634 times
635 }