]> git.proxmox.com Git - proxmox-backup.git/blob - src/pxar/sequential_decoder.rs
0ace15a902240fa7d20fa3c0f33a1e6ae3f038c5
[proxmox-backup.git] / src / pxar / sequential_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 SequentialDecoder<'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> SequentialDecoder<'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 pub (crate) fn get_reader_mut(&mut self) -> & mut R {
40 self.reader
41 }
42
43 pub (crate) fn read_item<T: Endian>(&mut self) -> Result<T, Error> {
44
45 let mut result: T = unsafe { std::mem::uninitialized() };
46
47 let buffer = unsafe { std::slice::from_raw_parts_mut(
48 &mut result as *mut T as *mut u8,
49 std::mem::size_of::<T>()
50 )};
51
52 self.reader.read_exact(buffer)?;
53
54 Ok(result.from_le())
55 }
56
57 fn read_symlink(&mut self, size: u64) -> Result<PathBuf, Error> {
58 if size < (HEADER_SIZE + 2) {
59 bail!("dectected short symlink target.");
60 }
61 let target_len = size - HEADER_SIZE;
62
63 if target_len > (libc::PATH_MAX as u64) {
64 bail!("symlink target too long ({}).", target_len);
65 }
66
67 let mut buffer = vec![0u8; target_len as usize];
68 self.reader.read_exact(&mut buffer)?;
69
70 let last_byte = buffer.pop().unwrap();
71 if last_byte != 0u8 {
72 bail!("symlink target not nul terminated.");
73 }
74
75 Ok(PathBuf::from(std::ffi::OsString::from_vec(buffer)))
76 }
77
78 pub (crate) fn read_filename(&mut self, size: u64) -> Result<OsString, Error> {
79 if size < (HEADER_SIZE + 2) {
80 bail!("dectected short filename");
81 }
82 let name_len = size - HEADER_SIZE;
83
84 if name_len > ((libc::FILENAME_MAX as u64) + 1) {
85 bail!("filename too long ({}).", name_len);
86 }
87
88 let mut buffer = vec![0u8; name_len as usize];
89 self.reader.read_exact(&mut buffer)?;
90
91 let last_byte = buffer.pop().unwrap();
92 if last_byte != 0u8 {
93 bail!("filename entry not nul terminated.");
94 }
95
96 if (buffer.len() == 1 && buffer[0] == b'.') || (buffer.len() == 2 && buffer[0] == b'.' && buffer[1] == b'.') {
97 bail!("found invalid filename with slashes.");
98 }
99
100 if buffer.iter().find(|b| (**b == b'/')).is_some() {
101 bail!("found invalid filename with slashes.");
102 }
103
104 let name = std::ffi::OsString::from_vec(buffer);
105 if name.is_empty() {
106 bail!("found empty filename.");
107 }
108
109 Ok(name)
110 }
111
112 fn restore_attributes(&mut self, _entry: &CaFormatEntry) -> Result<CaFormatHeader, Error> {
113
114 loop {
115 let head: CaFormatHeader = self.read_item()?;
116 match head.htype {
117 // fimxe: impl ...
118 _ => return Ok(head),
119 }
120 }
121 }
122
123 fn restore_mode(&mut self, entry: &CaFormatEntry, fd: RawFd) -> Result<(), Error> {
124
125 let mode = Mode::from_bits_truncate((entry.mode as u32) & 0o7777);
126
127 nix::sys::stat::fchmod(fd, mode)?;
128
129 Ok(())
130 }
131
132 fn restore_mode_at(&mut self, entry: &CaFormatEntry, dirfd: RawFd, filename: &OsStr) -> Result<(), Error> {
133
134 let mode = Mode::from_bits_truncate((entry.mode as u32) & 0o7777);
135
136 // NOTE: we want :FchmodatFlags::NoFollowSymlink, but fchmodat does not support that
137 // on linux (see man fchmodat). Fortunately, we can simply avoid calling this on symlinks.
138 nix::sys::stat::fchmodat(Some(dirfd), filename, mode, nix::sys::stat::FchmodatFlags::FollowSymlink)?;
139
140 Ok(())
141 }
142
143 fn restore_ugid(&mut self, entry: &CaFormatEntry, fd: RawFd) -> Result<(), Error> {
144
145 let uid = entry.uid as u32;
146 let gid = entry.gid as u32;
147
148 let res = unsafe { libc::fchown(fd, uid, gid) };
149 Errno::result(res)?;
150
151 Ok(())
152 }
153
154 fn restore_ugid_at(&mut self, entry: &CaFormatEntry, dirfd: RawFd, filename: &OsStr) -> Result<(), Error> {
155
156 let uid = entry.uid as u32;
157 let gid = entry.gid as u32;
158
159 let res = filename.with_nix_path(|cstr| unsafe {
160 libc::fchownat(dirfd, cstr.as_ptr(), uid, gid, libc::AT_SYMLINK_NOFOLLOW)
161 })?;
162 Errno::result(res)?;
163
164 Ok(())
165 }
166
167 fn restore_mtime(&mut self, entry: &CaFormatEntry, fd: RawFd) -> Result<(), Error> {
168
169 let times = nsec_to_update_timespec(entry.mtime);
170
171 let res = unsafe { libc::futimens(fd, &times[0]) };
172 Errno::result(res)?;
173
174 Ok(())
175 }
176
177 fn restore_mtime_at(&mut self, entry: &CaFormatEntry, dirfd: RawFd, filename: &OsStr) -> Result<(), Error> {
178
179 let times = nsec_to_update_timespec(entry.mtime);
180
181 let res = filename.with_nix_path(|cstr| unsafe {
182 libc::utimensat(dirfd, cstr.as_ptr(), &times[0], libc::AT_SYMLINK_NOFOLLOW)
183 })?;
184 Errno::result(res)?;
185
186 Ok(())
187 }
188
189 fn restore_device_at(&mut self, entry: &CaFormatEntry, dirfd: RawFd, filename: &OsStr, device: &CaFormatDevice) -> Result<(), Error> {
190
191 let rdev = nix::sys::stat::makedev(device.major, device.minor);
192 let mode = ((entry.mode as u32) & libc::S_IFMT) | 0o0600;
193 let res = filename.with_nix_path(|cstr| unsafe {
194 libc::mknodat(dirfd, cstr.as_ptr(), mode, rdev)
195 })?;
196 Errno::result(res)?;
197
198 Ok(())
199 }
200
201 fn restore_socket_at(&mut self, dirfd: RawFd, filename: &OsStr) -> Result<(), Error> {
202
203 let mode = libc::S_IFSOCK | 0o0600;
204 let res = filename.with_nix_path(|cstr| unsafe {
205 libc::mknodat(dirfd, cstr.as_ptr(), mode, 0)
206 })?;
207 Errno::result(res)?;
208
209 Ok(())
210 }
211
212 fn restore_fifo_at(&mut self, dirfd: RawFd, filename: &OsStr) -> Result<(), Error> {
213
214 let mode = libc::S_IFIFO | 0o0600;
215 let res = filename.with_nix_path(|cstr| unsafe {
216 libc::mkfifoat(dirfd, cstr.as_ptr(), mode)
217 })?;
218 Errno::result(res)?;
219
220 Ok(())
221 }
222
223 fn skip_bytes(&mut self, count: usize) -> Result<(), Error> {
224 let mut done = 0;
225 while done < count {
226 let todo = count - done;
227 let n = if todo > self.skip_buffer.len() { self.skip_buffer.len() } else { todo };
228 let data = &mut self.skip_buffer[..n];
229 self.reader.read_exact(data)?;
230 done += n;
231 }
232 Ok(())
233 }
234
235 /// Restore an archive into the specified directory.
236 ///
237 /// The directory is created if it does not exist.
238 pub fn restore<F>(
239 &mut self,
240 path: &Path,
241 callback: &F,
242 ) -> Result<(), Error>
243 where F: Fn(&Path) -> Result<(), Error>
244 {
245
246 let _ = std::fs::create_dir(path);
247
248 let dir = match nix::dir::Dir::open(path, nix::fcntl::OFlag::O_DIRECTORY, nix::sys::stat::Mode::empty()) {
249 Ok(dir) => dir,
250 Err(err) => bail!("unable to open target directory {:?} - {}", path, err),
251 };
252
253 self.restore_sequential(&mut path.to_owned(), &OsString::new(), &dir, callback)
254 }
255
256 fn restore_sequential<F>(
257 &mut self,
258 path: &mut PathBuf, // used for error reporting
259 filename: &OsStr, // repeats path last component
260 parent: &nix::dir::Dir,
261 callback: &F,
262 ) -> Result<(), Error>
263 where F: Fn(&Path) -> Result<(), Error>
264 {
265
266 let parent_fd = parent.as_raw_fd();
267
268 // read ENTRY first
269 let head: CaFormatHeader = self.read_item()?;
270 check_ca_header::<CaFormatEntry>(&head, CA_FORMAT_ENTRY)?;
271 let entry: CaFormatEntry = self.read_item()?;
272
273 (callback)(path)?;
274
275 let mode = entry.mode as u32; //fixme: upper 32bits?
276
277 let ifmt = mode & libc::S_IFMT;
278
279 if ifmt == libc::S_IFDIR {
280 let dir;
281 if filename.is_empty() {
282 dir = nix::dir::Dir::openat(parent_fd, ".", OFlag::O_DIRECTORY, Mode::empty())?;
283 } else {
284 dir = match dir_mkdirat(parent_fd, filename, true) {
285 Ok(dir) => dir,
286 Err(err) => bail!("unable to open directory {:?} - {}", path, err),
287 };
288 }
289
290 let mut head = self.restore_attributes(&entry)?;
291
292 while head.htype == CA_FORMAT_FILENAME {
293 let name = self.read_filename(head.size)?;
294 path.push(&name);
295 //println!("NAME: {:?}", path);
296 self.restore_sequential(path, &name, &dir, callback)?;
297 path.pop();
298
299 head = self.read_item()?;
300 }
301
302 if head.htype != CA_FORMAT_GOODBYE {
303 bail!("got unknown header type inside directory entry {:016x}", head.htype);
304 }
305
306 //println!("Skip Goodbye");
307 if head.size < HEADER_SIZE { bail!("detected short goodbye table"); }
308
309 self.skip_bytes((head.size - HEADER_SIZE) as usize)?;
310
311 self.restore_mode(&entry, dir.as_raw_fd())?;
312 self.restore_mtime(&entry, dir.as_raw_fd())?;
313 self.restore_ugid(&entry, dir.as_raw_fd())?;
314
315 return Ok(());
316 }
317
318 if filename.is_empty() {
319 bail!("got empty file name at {:?}", path)
320 }
321
322 if ifmt == libc::S_IFLNK {
323 // fixme: create symlink
324 //fixme: restore permission, acls, xattr, ...
325
326 let head: CaFormatHeader = self.read_item()?;
327 match head.htype {
328 CA_FORMAT_SYMLINK => {
329 let target = self.read_symlink(head.size)?;
330 //println!("TARGET: {:?}", target);
331 if let Err(err) = symlinkat(&target, parent_fd, filename) {
332 bail!("create symlink {:?} failed - {}", path, err);
333 }
334 }
335 _ => {
336 bail!("got unknown header type inside symlink entry {:016x}", head.htype);
337 }
338 }
339
340 // self.restore_mode_at(&entry, parent_fd, filename)?; //not supported on symlinks
341 self.restore_ugid_at(&entry, parent_fd, filename)?;
342 self.restore_mtime_at(&entry, parent_fd, filename)?;
343
344 return Ok(());
345 }
346
347 if ifmt == libc::S_IFSOCK {
348
349 self.restore_socket_at(parent_fd, filename)?;
350
351 self.restore_mode_at(&entry, parent_fd, filename)?;
352 self.restore_ugid_at(&entry, parent_fd, filename)?;
353 self.restore_mtime_at(&entry, parent_fd, filename)?;
354
355 return Ok(());
356 }
357
358 if ifmt == libc::S_IFIFO {
359
360 self.restore_fifo_at(parent_fd, filename)?;
361
362 self.restore_mode_at(&entry, parent_fd, filename)?;
363 self.restore_ugid_at(&entry, parent_fd, filename)?;
364 self.restore_mtime_at(&entry, parent_fd, filename)?;
365
366 return Ok(());
367 }
368
369 if (ifmt == libc::S_IFBLK) || (ifmt == libc::S_IFCHR) {
370
371 let head: CaFormatHeader = self.read_item()?;
372 match head.htype {
373 CA_FORMAT_DEVICE => {
374 let device: CaFormatDevice = self.read_item()?;
375 self.restore_device_at(&entry, parent_fd, filename, &device)?;
376 }
377 _ => {
378 bail!("got unknown header type inside device entry {:016x}", head.htype);
379 }
380 }
381
382 self.restore_mode_at(&entry, parent_fd, filename)?;
383 self.restore_ugid_at(&entry, parent_fd, filename)?;
384 self.restore_mtime_at(&entry, parent_fd, filename)?;
385
386 return Ok(());
387 }
388
389 if ifmt == libc::S_IFREG {
390
391 let mut read_buffer: [u8; 64*1024] = unsafe { std::mem::uninitialized() };
392
393 let flags = OFlag::O_CREAT|OFlag::O_WRONLY|OFlag::O_EXCL;
394 let open_mode = Mode::from_bits_truncate(0o0600 | mode);
395
396 let mut file = match file_openat(parent_fd, filename, flags, open_mode) {
397 Ok(file) => file,
398 Err(err) => bail!("open file {:?} failed - {}", path, err),
399 };
400
401 let head = self.restore_attributes(&entry)?;
402
403 if head.htype != CA_FORMAT_PAYLOAD {
404 bail!("got unknown header type for file entry {:016x}", head.htype);
405 }
406
407 if head.size < HEADER_SIZE {
408 bail!("detected short payload");
409 }
410 let need = (head.size - HEADER_SIZE) as usize;
411 //self.reader.seek(SeekFrom::Current(need as i64))?;
412
413 let mut done = 0;
414 while done < need {
415 let todo = need - done;
416 let n = if todo > read_buffer.len() { read_buffer.len() } else { todo };
417 let data = &mut read_buffer[..n];
418 self.reader.read_exact(data)?;
419 file.write_all(data)?;
420 done += n;
421 }
422
423 self.restore_mode(&entry, file.as_raw_fd())?;
424 self.restore_mtime(&entry, file.as_raw_fd())?;
425 self.restore_ugid(&entry, file.as_raw_fd())?;
426
427 return Ok(());
428 }
429
430 Ok(())
431 }
432
433 /// List/Dump archive content.
434 ///
435 /// Simply print the list of contained files. This dumps archive
436 /// format details when the verbose flag is set (useful for debug).
437 pub fn dump_entry<W: std::io::Write>(
438 &mut self,
439 path: &mut PathBuf,
440 verbose: bool,
441 output: &mut W,
442 ) -> Result<(), Error> {
443
444 let print_head = |head: &CaFormatHeader| {
445 println!("Type: {:016x}", head.htype);
446 println!("Size: {}", head.size);
447 };
448
449 let head: CaFormatHeader = self.read_item()?;
450 if verbose {
451 println!("Path: {:?}", path);
452 print_head(&head);
453 } else {
454 println!("{:?}", path);
455 }
456
457 check_ca_header::<CaFormatEntry>(&head, CA_FORMAT_ENTRY)?;
458 let entry: CaFormatEntry = self.read_item()?;
459
460 if verbose {
461 println!("Mode: {:08x} {:08x}", entry.mode, (entry.mode as u32) & libc::S_IFDIR);
462 }
463 // fixme: dump attributes (ACLs, ...)
464
465 let ifmt = (entry.mode as u32) & libc::S_IFMT;
466
467 if ifmt == libc::S_IFDIR {
468
469 let mut entry_count = 0;
470
471 loop {
472 let head: CaFormatHeader = self.read_item()?;
473 if verbose {
474 print_head(&head);
475 }
476 match head.htype {
477
478 CA_FORMAT_FILENAME => {
479 let name = self.read_filename(head.size)?;
480 if verbose { println!("Name: {:?}", name); }
481 entry_count += 1;
482 path.push(&name);
483 self.dump_entry(path, verbose, output)?;
484 path.pop();
485 }
486 CA_FORMAT_GOODBYE => {
487 let table_size = (head.size - HEADER_SIZE) as usize;
488 if verbose {
489 println!("Goodbye: {:?}", path);
490 self.dump_goodby_entries(entry_count, table_size)?;
491 } else {
492 self.skip_bytes(table_size)?;
493 }
494 break;
495 }
496 _ => {
497 panic!("got unexpected header type inside directory");
498 }
499 }
500 }
501 } else {
502
503 let head: CaFormatHeader = self.read_item()?;
504 if verbose {
505 print_head(&head);
506 }
507
508 match head.htype {
509
510 CA_FORMAT_SYMLINK => {
511 let target = self.read_symlink(head.size)?;
512 if verbose {
513 println!("Symlink: {:?}", target);
514 }
515 }
516 CA_FORMAT_DEVICE => {
517 let device: CaFormatDevice = self.read_item()?;
518 if verbose {
519 println!("Device: {}, {}", device.major, device.minor);
520 }
521 }
522 CA_FORMAT_PAYLOAD => {
523 let payload_size = (head.size - HEADER_SIZE) as usize;
524 if verbose {
525 println!("Payload: {}", payload_size);
526 }
527 self.skip_bytes(payload_size)?;
528 }
529 _ => {
530 panic!("got unexpected header type inside non-directory");
531 }
532 }
533 }
534
535 Ok(())
536 }
537
538 fn dump_goodby_entries(
539 &mut self,
540 entry_count: usize,
541 table_size: usize,
542 ) -> Result<(), Error> {
543
544 let item_size = std::mem::size_of::<CaFormatGoodbyeItem>();
545 if table_size < item_size {
546 bail!("Goodbye table to small ({} < {})", table_size, item_size);
547 }
548 if (table_size % item_size) != 0 {
549 bail!("Goodbye table with strange size ({})", table_size);
550 }
551
552 let entries = table_size / item_size;
553
554 if entry_count != (entries - 1) {
555 bail!("Goodbye table with wrong entry count ({} != {})", entry_count, entries - 1);
556 }
557
558 let mut count = 0;
559
560 loop {
561 let item: CaFormatGoodbyeItem = self.read_item()?;
562 count += 1;
563 if item.hash == CA_FORMAT_GOODBYE_TAIL_MARKER {
564 if count != entries {
565 bail!("unexpected goodbye tail marker");
566 }
567 println!("Goodby tail mark.");
568 break;
569 }
570 println!("Goodby item: offset {}, size {}, hash {:016x}", item.offset, item.size, item.hash);
571 if count >= (table_size / item_size) {
572 bail!("too many goodbye items (no tail marker)");
573 }
574 }
575
576 Ok(())
577 }
578 }
579
580 fn file_openat(parent: RawFd, filename: &OsStr, flags: OFlag, mode: Mode) -> Result<std::fs::File, Error> {
581
582 let fd = filename.with_nix_path(|cstr| {
583 nix::fcntl::openat(parent, cstr.as_ref(), flags, mode)
584 })??;
585
586 let file = unsafe { std::fs::File::from_raw_fd(fd) };
587
588 Ok(file)
589 }
590
591 fn dir_mkdirat(parent: RawFd, filename: &OsStr, create_new: bool) -> Result<nix::dir::Dir, nix::Error> {
592
593 // call mkdirat first
594 let res = filename.with_nix_path(|cstr| unsafe {
595 libc::mkdirat(parent, cstr.as_ptr(), libc::S_IRWXU)
596 })?;
597
598 match Errno::result(res) {
599 Ok(_) => {},
600 Err(err) => {
601 if err == nix::Error::Sys(nix::errno::Errno::EEXIST) {
602 if create_new { return Err(err); }
603 } else {
604 return Err(err);
605 }
606 }
607 }
608
609 let dir = nix::dir::Dir::openat(parent, filename, OFlag::O_DIRECTORY, Mode::empty())?;
610
611 Ok(dir)
612 }
613
614 fn symlinkat(target: &Path, parent: RawFd, linkname: &OsStr) -> Result<(), Error> {
615
616 target.with_nix_path(|target| {
617 linkname.with_nix_path(|linkname| {
618 let res = unsafe { libc::symlinkat(target.as_ptr(), parent, linkname.as_ptr()) };
619 Errno::result(res)?;
620 Ok(())
621 })?
622 })?
623 }
624
625 fn nsec_to_update_timespec(mtime_nsec: u64) -> [libc::timespec; 2] {
626
627 // restore mtime
628 const UTIME_OMIT: i64 = ((1 << 30) - 2);
629 const NANOS_PER_SEC: i64 = 1_000_000_000;
630
631 let sec = (mtime_nsec as i64) / NANOS_PER_SEC;
632 let nsec = (mtime_nsec as i64) % NANOS_PER_SEC;
633
634 let times: [libc::timespec; 2] = [
635 libc::timespec { tv_sec: 0, tv_nsec: UTIME_OMIT },
636 libc::timespec { tv_sec: sec, tv_nsec: nsec },
637 ];
638
639 times
640 }