]> git.proxmox.com Git - proxmox-backup.git/blob - src/pxar/encoder.rs
src/pxar/encoder.rs: add new skip_lost_and_found parameter
[proxmox-backup.git] / src / pxar / encoder.rs
1 //! *pxar* format encoder.
2 //!
3 //! This module contain the code to generate *pxar* archive files.
4
5 use failure::*;
6 use endian_trait::Endian;
7 use std::collections::{HashSet, HashMap};
8
9 use super::format_definition::*;
10 use super::binary_search_tree::*;
11 use super::helper::*;
12 use super::exclude_pattern::*;
13 use crate::tools::fs;
14 use crate::tools::acl;
15 use crate::tools::xattr;
16
17 use std::io::Write;
18 use std::os::unix::io::AsRawFd;
19 use std::os::unix::ffi::OsStrExt;
20 use std::os::unix::io::RawFd;
21 use std::path::{Path, PathBuf};
22
23 use std::ffi::CStr;
24
25 use nix::NixPath;
26 use nix::fcntl::OFlag;
27 use nix::sys::stat::Mode;
28 use nix::errno::Errno;
29 use nix::sys::stat::FileStat;
30
31 use proxmox::tools::vec;
32
33 /// The format requires to build sorted directory lookup tables in
34 /// memory, so we restrict the number of allowed entries to limit
35 /// maximum memory usage.
36 pub const MAX_DIRECTORY_ENTRIES: usize = 256*1024;
37
38 #[derive(Eq, PartialEq, Hash)]
39 struct HardLinkInfo {
40 st_dev: u64,
41 st_ino: u64,
42 }
43
44 pub struct Encoder<'a, W: Write> {
45 base_path: PathBuf,
46 relative_path: PathBuf,
47 writer: &'a mut W,
48 writer_pos: usize,
49 _size: usize,
50 file_copy_buffer: Vec<u8>,
51 device_set: Option<HashSet<u64>>,
52 verbose: bool,
53 // Flags set by the user
54 feature_flags: u64,
55 // Flags signaling features supported by the filesystem
56 fs_feature_flags: u64,
57 hardlinks: HashMap<HardLinkInfo, (PathBuf, u64)>,
58 }
59
60 impl <'a, W: Write> Encoder<'a, W> {
61
62 // used for error reporting
63 fn full_path(&self) -> PathBuf {
64 self.base_path.join(&self.relative_path)
65 }
66
67 /// Create archive, write result data to ``writer``.
68 ///
69 /// The ``device_set`` can be use used to limit included mount points.
70 ///
71 /// - ``None``: include all mount points
72 /// - ``Some(set)``: only include devices listed in this set (the
73 /// root path device is automathically added to this list, so
74 /// you can pass an empty set if you want to archive a single
75 /// mount point.)
76 pub fn encode(
77 path: PathBuf,
78 dir: &mut nix::dir::Dir,
79 writer: &'a mut W,
80 device_set: Option<HashSet<u64>>,
81 verbose: bool,
82 skip_lost_and_found: bool, // fixme: should be a feature flag ??
83 feature_flags: u64,
84 ) -> Result<(), Error> {
85
86 const FILE_COPY_BUFFER_SIZE: usize = 1024*1024;
87
88 let mut file_copy_buffer = Vec::with_capacity(FILE_COPY_BUFFER_SIZE);
89 unsafe { file_copy_buffer.set_len(FILE_COPY_BUFFER_SIZE); }
90
91
92 // todo: use scandirat??
93
94 let dir_fd = dir.as_raw_fd();
95 let stat = nix::sys::stat::fstat(dir_fd)
96 .map_err(|err| format_err!("fstat {:?} failed - {}", path, err))?;
97
98 if !is_directory(&stat) {
99 bail!("got unexpected file type {:?} (not a directory)", path);
100 }
101
102 let mut device_set = device_set.clone();
103 if let Some(ref mut set) = device_set {
104 set.insert(stat.st_dev);
105 }
106
107 let magic = detect_fs_type(dir_fd)?;
108
109 if is_virtual_file_system(magic) {
110 bail!("backup virtual file systems is disabled!");
111 }
112
113 let fs_feature_flags = feature_flags_from_magic(magic);
114
115 let mut me = Self {
116 base_path: path,
117 relative_path: PathBuf::new(),
118 writer: writer,
119 writer_pos: 0,
120 _size: 0,
121 file_copy_buffer,
122 device_set,
123 verbose,
124 feature_flags,
125 fs_feature_flags,
126 hardlinks: HashMap::new(),
127 };
128
129 if verbose { println!("{:?}", me.full_path()); }
130
131 let mut excludes = Vec::new();
132 if skip_lost_and_found {
133 excludes.push(PxarExcludePattern::from_line(b"**/lost+found").unwrap().unwrap());
134 }
135 me.encode_dir(dir, &stat, magic, excludes)?;
136
137 Ok(())
138 }
139
140 fn write(&mut self, buf: &[u8]) -> Result<(), Error> {
141 self.writer.write_all(buf)?;
142 self.writer_pos += buf.len();
143 Ok(())
144 }
145
146 fn write_item<T: Endian>(&mut self, item: T) -> Result<(), Error> {
147
148 let data = item.to_le();
149
150 let buffer = unsafe { std::slice::from_raw_parts(
151 &data as *const T as *const u8,
152 std::mem::size_of::<T>()
153 )};
154
155 self.write(buffer)?;
156
157 Ok(())
158 }
159
160 fn flush_copy_buffer(&mut self, size: usize) -> Result<(), Error> {
161 self.writer.write_all(&self.file_copy_buffer[..size])?;
162 self.writer_pos += size;
163 Ok(())
164 }
165
166 fn write_header(&mut self, htype: u64, size: u64) -> Result<(), Error> {
167
168 let size = size + (std::mem::size_of::<CaFormatHeader>() as u64);
169 self.write_item(CaFormatHeader { size, htype })?;
170
171 Ok(())
172 }
173
174 fn write_filename(&mut self, name: &CStr) -> Result<(), Error> {
175
176 let buffer = name.to_bytes_with_nul();
177 self.write_header(CA_FORMAT_FILENAME, buffer.len() as u64)?;
178 self.write(buffer)?;
179
180 Ok(())
181 }
182
183 fn create_entry(&self, stat: &FileStat) -> Result<CaFormatEntry, Error> {
184
185 let mode = if is_symlink(&stat) {
186 (libc::S_IFLNK | 0o777) as u64
187 } else {
188 (stat.st_mode & (libc::S_IFMT | 0o7777)) as u64
189 };
190
191 let mtime = stat.st_mtime * 1_000_000_000 + stat.st_mtime_nsec;
192 if mtime < 0 {
193 bail!("got strange mtime ({}) from fstat for {:?}.", mtime, self.full_path());
194 }
195
196
197 let entry = CaFormatEntry {
198 mode: mode,
199 flags: 0,
200 uid: stat.st_uid as u64,
201 gid: stat.st_gid as u64,
202 mtime: mtime as u64,
203 };
204
205 Ok(entry)
206 }
207
208 fn read_chattr(&self, fd: RawFd, entry: &mut CaFormatEntry) -> Result<(), Error> {
209
210 let mut attr: usize = 0;
211
212 let res = unsafe { fs::read_attr_fd(fd, &mut attr)};
213 if let Err(err) = res {
214 if let nix::Error::Sys(errno) = err {
215 if errno_is_unsupported(errno) { return Ok(()) };
216 }
217 bail!("read_attr_fd failed for {:?} - {}", self.full_path(), err);
218 }
219
220 let flags = ca_feature_flags_from_chattr(attr as u32);
221 entry.flags = entry.flags | flags;
222
223 Ok(())
224 }
225
226 fn read_fat_attr(&self, fd: RawFd, magic: i64, entry: &mut CaFormatEntry) -> Result<(), Error> {
227
228 if magic != MSDOS_SUPER_MAGIC && magic != FUSE_SUPER_MAGIC { return Ok(()); }
229
230 let mut attr: u32 = 0;
231
232 let res = unsafe { fs::read_fat_attr_fd(fd, &mut attr)};
233 if let Err(err) = res {
234 if let nix::Error::Sys(errno) = err {
235 if errno_is_unsupported(errno) { return Ok(()) };
236 }
237 bail!("read_fat_attr_fd failed for {:?} - {}", self.full_path(), err);
238 }
239
240 let flags = ca_feature_flags_from_fat_attr(attr);
241 entry.flags = entry.flags | flags;
242
243 Ok(())
244 }
245
246 /// True if all of the given feature flags are set in the Encoder, false otherwise
247 fn has_features(&self, feature_flags: u64) -> bool {
248 (self.feature_flags & self.fs_feature_flags & feature_flags) == feature_flags
249 }
250
251 /// True if at least one of the given feature flags is set in the Encoder, false otherwise
252 fn has_some_features(&self, feature_flags: u64) -> bool {
253 (self.feature_flags & self.fs_feature_flags & feature_flags) != 0
254 }
255
256 fn read_xattrs(&self, fd: RawFd, stat: &FileStat) -> Result<(Vec<CaFormatXAttr>, Option<CaFormatFCaps>), Error> {
257 let mut xattrs = Vec::new();
258 let mut fcaps = None;
259
260 let flags = CA_FORMAT_WITH_XATTRS | CA_FORMAT_WITH_FCAPS;
261 if !self.has_some_features(flags) {
262 return Ok((xattrs, fcaps));
263 }
264 // Should never be called on symlinks, just in case check anyway
265 if is_symlink(&stat) {
266 return Ok((xattrs, fcaps));
267 }
268
269 let xattr_names = match xattr::flistxattr(fd) {
270 Ok(names) => names,
271 // Do not bail if the underlying endpoint does not supports xattrs
272 Err(Errno::EOPNOTSUPP) => return Ok((xattrs, fcaps)),
273 // Do not bail if the endpoint cannot carry xattrs (such as symlinks)
274 Err(Errno::EBADF) => return Ok((xattrs, fcaps)),
275 Err(err) => bail!("read_xattrs failed for {:?} - {}", self.full_path(), err),
276 };
277
278 for name in xattr_names.split(|c| *c == b'\0') {
279 // Only extract the relevant extended attributes
280 if !xattr::is_valid_xattr_name(&name) {
281 continue;
282 }
283
284 let value = match xattr::fgetxattr(fd, name) {
285 Ok(value) => value,
286 // Vanished between flistattr and getxattr, this is ok, silently ignore
287 Err(Errno::ENODATA) => continue,
288 Err(err) => bail!("read_xattrs failed for {:?} - {}", self.full_path(), err),
289 };
290
291 if xattr::is_security_capability(&name) {
292 if self.has_features(CA_FORMAT_WITH_FCAPS) {
293 // fcaps are stored in own format within the archive
294 fcaps = Some(CaFormatFCaps {
295 data: value,
296 });
297 }
298 } else if self.has_features(CA_FORMAT_WITH_XATTRS) {
299 xattrs.push(CaFormatXAttr {
300 name: name.to_vec(),
301 value: value,
302 });
303 }
304 }
305 xattrs.sort();
306
307 Ok((xattrs, fcaps))
308 }
309
310 fn read_acl(&self, fd: RawFd, stat: &FileStat, acl_type: acl::ACLType) -> Result<PxarACL, Error> {
311 let ret = PxarACL {
312 users: Vec::new(),
313 groups: Vec::new(),
314 group_obj: None,
315 default: None,
316 };
317
318 if !self.has_features(CA_FORMAT_WITH_ACL) {
319 return Ok(ret);
320 }
321 if is_symlink(&stat) {
322 return Ok(ret);
323 }
324 if acl_type == acl::ACL_TYPE_DEFAULT && !is_directory(&stat) {
325 bail!("ACL_TYPE_DEFAULT only defined for directories.");
326 }
327
328 // In order to be able to get ACLs with type ACL_TYPE_DEFAULT, we have
329 // to create a path for acl_get_file(). acl_get_fd() only allows to get
330 // ACL_TYPE_ACCESS attributes.
331 let proc_path = Path::new("/proc/self/fd/").join(fd.to_string());
332 let acl = match acl::ACL::get_file(&proc_path, acl_type) {
333 Ok(acl) => acl,
334 // Don't bail if underlying endpoint does not support acls
335 Err(Errno::EOPNOTSUPP) => return Ok(ret),
336 // Don't bail if the endpoint cannot carry acls
337 Err(Errno::EBADF) => return Ok(ret),
338 // Don't bail if there is no data
339 Err(Errno::ENODATA) => return Ok(ret),
340 Err(err) => bail!("error while reading ACL - {}", err),
341 };
342
343 self.process_acl(acl, acl_type)
344 }
345
346 fn process_acl(&self, acl: acl::ACL, acl_type: acl::ACLType) -> Result<PxarACL, Error> {
347 let mut acl_user = Vec::new();
348 let mut acl_group = Vec::new();
349 let mut acl_group_obj = None;
350 let mut acl_default = None;
351 let mut user_obj_permissions = None;
352 let mut group_obj_permissions = None;
353 let mut other_permissions = None;
354 let mut mask_permissions = None;
355
356 for entry in &mut acl.entries() {
357 let tag = entry.get_tag_type()?;
358 let permissions = entry.get_permissions()?;
359 match tag {
360 acl::ACL_USER_OBJ => user_obj_permissions = Some(permissions),
361 acl::ACL_GROUP_OBJ => group_obj_permissions = Some(permissions),
362 acl::ACL_OTHER => other_permissions = Some(permissions),
363 acl::ACL_MASK => mask_permissions = Some(permissions),
364 acl::ACL_USER => {
365 acl_user.push(CaFormatACLUser {
366 uid: entry.get_qualifier()?,
367 permissions: permissions,
368 });
369 },
370 acl::ACL_GROUP => {
371 acl_group.push(CaFormatACLGroup {
372 gid: entry.get_qualifier()?,
373 permissions: permissions,
374 });
375 },
376 _ => bail!("Unexpected ACL tag encountered!"),
377 }
378 }
379
380 acl_user.sort();
381 acl_group.sort();
382
383 match acl_type {
384 acl::ACL_TYPE_ACCESS => {
385 // The mask permissions are mapped to the stat group permissions
386 // in case that the ACL group permissions were set.
387 // Only in that case we need to store the group permissions,
388 // in the other cases they are identical to the stat group permissions.
389 if let (Some(gop), Some(_)) = (group_obj_permissions, mask_permissions) {
390 acl_group_obj = Some(CaFormatACLGroupObj {
391 permissions: gop,
392 });
393 }
394 },
395 acl::ACL_TYPE_DEFAULT => {
396 if user_obj_permissions != None ||
397 group_obj_permissions != None ||
398 other_permissions != None ||
399 mask_permissions != None
400 {
401 acl_default = Some(CaFormatACLDefault {
402 // The value is set to UINT64_MAX as placeholder if one
403 // of the permissions is not set
404 user_obj_permissions: user_obj_permissions.unwrap_or(std::u64::MAX),
405 group_obj_permissions: group_obj_permissions.unwrap_or(std::u64::MAX),
406 other_permissions: other_permissions.unwrap_or(std::u64::MAX),
407 mask_permissions: mask_permissions.unwrap_or(std::u64::MAX),
408 });
409 }
410 },
411 _ => bail!("Unexpected ACL type encountered"),
412 }
413
414 Ok(PxarACL {
415 users: acl_user,
416 groups: acl_group,
417 group_obj: acl_group_obj,
418 default: acl_default,
419 })
420 }
421
422 /// Read the project quota id for an inode, supported on ext4/XFS/FUSE/(ZFS TODO impl) filesystems
423 fn read_quota_project_id(&self, fd: RawFd, magic: i64, stat: &FileStat) -> Result<Option<CaFormatQuotaProjID>, Error> {
424 if !(is_directory(&stat) || is_reg_file(&stat)) {
425 return Ok(None);
426 }
427 if !self.has_features(CA_FORMAT_WITH_QUOTA_PROJID) {
428 return Ok(None);
429 }
430
431 match magic {
432 //TODO ZFS quota
433 EXT4_SUPER_MAGIC | XFS_SUPER_MAGIC | FUSE_SUPER_MAGIC => {
434 let mut fsxattr = fs::FSXAttr::default();
435 let res = unsafe {
436 fs::fs_ioc_fsgetxattr(fd, &mut fsxattr)
437 };
438
439 // On some FUSE filesystems it can happen that ioctl is not supported.
440 // For these cases projid is set to 0 while the error is ignored.
441 if let Err(err) = res {
442 let errno = err.as_errno().ok_or_else(|| {
443 format_err!("error while reading quota project id for {:#?}", self.full_path())
444 })?;
445 if errno_is_unsupported(errno) {
446 return Ok(None);
447 } else {
448 bail!("error while reading quota project id for {:#?} - {}", self.full_path(), errno);
449 }
450 }
451
452 let projid = fsxattr.fsx_projid as u64;
453 if projid == 0 {
454 return Ok(None);
455 } else {
456 return Ok(Some(CaFormatQuotaProjID { projid }));
457 }
458 },
459 _ => return Ok(None),
460 }
461 }
462
463 fn write_entry(&mut self, entry: CaFormatEntry) -> Result<(), Error> {
464
465 self.write_header(CA_FORMAT_ENTRY, std::mem::size_of::<CaFormatEntry>() as u64)?;
466 self.write_item(entry)?;
467
468 Ok(())
469 }
470
471 fn write_xattr(&mut self, xattr: CaFormatXAttr) -> Result<(), Error> {
472 let size = xattr.name.len() + xattr.value.len() + 1; // +1 for '\0' separating name and value
473 self.write_header(CA_FORMAT_XATTR, size as u64)?;
474 self.write(xattr.name.as_slice())?;
475 self.write(&[0])?;
476 self.write(xattr.value.as_slice())?;
477
478 Ok(())
479 }
480
481 fn write_fcaps(&mut self, fcaps: Option<CaFormatFCaps>) -> Result<(), Error> {
482 if let Some(fcaps) = fcaps {
483 let size = fcaps.data.len();
484 self.write_header(CA_FORMAT_FCAPS, size as u64)?;
485 self.write(fcaps.data.as_slice())?;
486 }
487
488 Ok(())
489 }
490
491 fn write_acl_user(&mut self, acl_user: CaFormatACLUser) -> Result<(), Error> {
492 self.write_header(CA_FORMAT_ACL_USER, std::mem::size_of::<CaFormatACLUser>() as u64)?;
493 self.write_item(acl_user)?;
494
495 Ok(())
496 }
497
498 fn write_acl_group(&mut self, acl_group: CaFormatACLGroup) -> Result<(), Error> {
499 self.write_header(CA_FORMAT_ACL_GROUP, std::mem::size_of::<CaFormatACLGroup>() as u64)?;
500 self.write_item(acl_group)?;
501
502 Ok(())
503 }
504
505 fn write_acl_group_obj(&mut self, acl_group_obj: CaFormatACLGroupObj) -> Result<(), Error> {
506 self.write_header(CA_FORMAT_ACL_GROUP_OBJ, std::mem::size_of::<CaFormatACLGroupObj>() as u64)?;
507 self.write_item(acl_group_obj)?;
508
509 Ok(())
510 }
511
512 fn write_acl_default(&mut self, acl_default: CaFormatACLDefault) -> Result<(), Error> {
513 self.write_header(CA_FORMAT_ACL_DEFAULT, std::mem::size_of::<CaFormatACLDefault>() as u64)?;
514 self.write_item(acl_default)?;
515
516 Ok(())
517 }
518
519 fn write_acl_default_user(&mut self, acl_default_user: CaFormatACLUser) -> Result<(), Error> {
520 self.write_header(CA_FORMAT_ACL_DEFAULT_USER, std::mem::size_of::<CaFormatACLUser>() as u64)?;
521 self.write_item(acl_default_user)?;
522
523 Ok(())
524 }
525
526 fn write_acl_default_group(&mut self, acl_default_group: CaFormatACLGroup) -> Result<(), Error> {
527 self.write_header(CA_FORMAT_ACL_DEFAULT_GROUP, std::mem::size_of::<CaFormatACLGroup>() as u64)?;
528 self.write_item(acl_default_group)?;
529
530 Ok(())
531 }
532
533 fn write_quota_project_id(&mut self, projid: CaFormatQuotaProjID) -> Result<(), Error> {
534 self.write_header(CA_FORMAT_QUOTA_PROJID, std::mem::size_of::<CaFormatQuotaProjID>() as u64)?;
535 self.write_item(projid)?;
536
537 Ok(())
538 }
539
540 fn write_goodbye_table(&mut self, goodbye_offset: usize, goodbye_items: &mut [CaFormatGoodbyeItem]) -> Result<(), Error> {
541
542 goodbye_items.sort_unstable_by(|a, b| a.hash.cmp(&b.hash));
543
544 let item_count = goodbye_items.len();
545
546 let goodbye_table_size = (item_count + 1)*std::mem::size_of::<CaFormatGoodbyeItem>();
547
548 self.write_header(CA_FORMAT_GOODBYE, goodbye_table_size as u64)?;
549
550 if self.file_copy_buffer.len() < goodbye_table_size {
551 let need = goodbye_table_size - self.file_copy_buffer.len();
552 self.file_copy_buffer.reserve(need);
553 unsafe { self.file_copy_buffer.set_len(self.file_copy_buffer.capacity()); }
554 }
555
556 let buffer = &mut self.file_copy_buffer;
557
558 copy_binary_search_tree(item_count, |s, d| {
559 let item = &goodbye_items[s];
560 let offset = d*std::mem::size_of::<CaFormatGoodbyeItem>();
561 let dest = crate::tools::map_struct_mut::<CaFormatGoodbyeItem>(&mut buffer[offset..]).unwrap();
562 dest.offset = u64::to_le(item.offset);
563 dest.size = u64::to_le(item.size);
564 dest.hash = u64::to_le(item.hash);
565 });
566
567 // append CaFormatGoodbyeTail as last item
568 let offset = item_count*std::mem::size_of::<CaFormatGoodbyeItem>();
569 let dest = crate::tools::map_struct_mut::<CaFormatGoodbyeItem>(&mut buffer[offset..]).unwrap();
570 dest.offset = u64::to_le(goodbye_offset as u64);
571 dest.size = u64::to_le((goodbye_table_size + std::mem::size_of::<CaFormatHeader>()) as u64);
572 dest.hash = u64::to_le(CA_FORMAT_GOODBYE_TAIL_MARKER);
573
574 self.flush_copy_buffer(goodbye_table_size)?;
575
576 Ok(())
577 }
578
579 fn encode_dir(&mut self, dir: &mut nix::dir::Dir, dir_stat: &FileStat, magic: i64, match_pattern: Vec<PxarExcludePattern>) -> Result<(), Error> {
580
581 //println!("encode_dir: {:?} start {}", self.full_path(), self.writer_pos);
582
583 let mut name_list = vec![];
584
585 let rawfd = dir.as_raw_fd();
586
587 let dir_start_pos = self.writer_pos;
588
589 let mut dir_entry = self.create_entry(&dir_stat)?;
590
591 self.read_chattr(rawfd, &mut dir_entry)?;
592 self.read_fat_attr(rawfd, magic, &mut dir_entry)?;
593
594 // for each node in the directory tree, the filesystem features are
595 // checked based on the fs magic number.
596 self.fs_feature_flags = feature_flags_from_magic(magic);
597
598 let (xattrs, fcaps) = self.read_xattrs(rawfd, &dir_stat)?;
599 let acl_access = self.read_acl(rawfd, &dir_stat, acl::ACL_TYPE_ACCESS)?;
600 let acl_default = self.read_acl(rawfd, &dir_stat, acl::ACL_TYPE_DEFAULT)?;
601 let projid = self.read_quota_project_id(rawfd, magic, &dir_stat)?;
602
603 self.write_entry(dir_entry)?;
604 for xattr in xattrs {
605 self.write_xattr(xattr)?;
606 }
607 self.write_fcaps(fcaps)?;
608
609 for user in acl_access.users {
610 self.write_acl_user(user)?;
611 }
612 for group in acl_access.groups {
613 self.write_acl_group(group)?;
614 }
615 if let Some(group_obj) = acl_access.group_obj {
616 self.write_acl_group_obj(group_obj)?;
617 }
618
619 for default_user in acl_default.users {
620 self.write_acl_default_user(default_user)?;
621 }
622 for default_group in acl_default.groups {
623 self.write_acl_default_group(default_group)?;
624 }
625 if let Some(default) = acl_default.default {
626 self.write_acl_default(default)?;
627 }
628 if let Some(projid) = projid {
629 self.write_quota_project_id(projid)?;
630 }
631
632 let include_children;
633 if is_virtual_file_system(magic) {
634 include_children = false;
635 } else {
636 if let Some(set) = &self.device_set {
637 include_children = set.contains(&dir_stat.st_dev);
638 } else {
639 include_children = true;
640 }
641 }
642
643 // Expand the exclude match pattern inherited from the parent by local entries, if present
644 let mut local_match_pattern = match_pattern.clone();
645 let pxar_exclude = match PxarExcludePattern::from_file(rawfd, ".pxarexclude") {
646 Ok(Some((mut excludes, buffer, stat))) => {
647 local_match_pattern.append(&mut excludes);
648 Some((buffer, stat))
649 },
650 Ok(None) => None,
651 Err(err) => bail!("error while reading exclude file - {}", err),
652 };
653
654 if include_children {
655 for entry in dir.iter() {
656 let entry = entry.map_err(|err| {
657 format_err!("readir {:?} failed - {}", self.full_path(), err)
658 })?;
659 let filename = entry.file_name().to_owned();
660
661 let name = filename.to_bytes_with_nul();
662 if name == b".\0" || name == b"..\0" {
663 continue;
664 }
665
666 let stat = match nix::sys::stat::fstatat(rawfd, filename.as_ref(), nix::fcntl::AtFlags::AT_SYMLINK_NOFOLLOW) {
667 Ok(stat) => stat,
668 Err(nix::Error::Sys(Errno::ENOENT)) => {
669 let filename_osstr = std::ffi::OsStr::from_bytes(filename.to_bytes());
670 self.report_vanished_file(&self.full_path().join(filename_osstr))?;
671 continue;
672 },
673 Err(err) => bail!("fstat {:?} failed - {}", self.full_path(), err),
674 };
675
676 match match_exclude_pattern(&filename, &stat, &local_match_pattern) {
677 (MatchType::Exclude, _) => {
678 let filename_osstr = std::ffi::OsStr::from_bytes(filename.to_bytes());
679 eprintln!("matched by .pxarexclude entry - skipping: {:?}", self.full_path().join(filename_osstr));
680 },
681 (_, child_pattern) => name_list.push((filename, stat, child_pattern)),
682 }
683
684 if name_list.len() > MAX_DIRECTORY_ENTRIES {
685 bail!("too many directory items in {:?} (> {})", self.full_path(), MAX_DIRECTORY_ENTRIES);
686 }
687 }
688 } else {
689 eprintln!("skip mount point: {:?}", self.full_path());
690 }
691
692 name_list.sort_unstable_by(|a, b| a.0.cmp(&b.0));
693
694 let mut goodbye_items = vec![];
695
696 for (filename, stat, exclude_list) in name_list {
697 let start_pos = self.writer_pos;
698
699 if filename.as_bytes() == b".pxarexclude" {
700 if let Some((ref content, ref stat)) = pxar_exclude {
701 let filefd = match nix::fcntl::openat(rawfd, filename.as_ref(), OFlag::O_NOFOLLOW, Mode::empty()) {
702 Ok(filefd) => filefd,
703 Err(nix::Error::Sys(Errno::ENOENT)) => {
704 self.report_vanished_file(&self.full_path())?;
705 continue;
706 },
707 Err(err) => {
708 let filename_osstr = std::ffi::OsStr::from_bytes(filename.to_bytes());
709 bail!("open file {:?} failed - {}", self.full_path().join(filename_osstr), err);
710 },
711 };
712
713 let child_magic = if dir_stat.st_dev != stat.st_dev {
714 detect_fs_type(filefd)?
715 } else {
716 magic
717 };
718
719 self.write_filename(&filename)?;
720 self.encode_pxar_exclude(filefd, stat, child_magic, content)?;
721 continue;
722 }
723 }
724
725 self.relative_path.push(std::ffi::OsStr::from_bytes(filename.as_bytes()));
726
727 if self.verbose { println!("{:?}", self.full_path()); }
728
729 if is_directory(&stat) {
730
731 let mut dir = match nix::dir::Dir::openat(rawfd, filename.as_ref(), OFlag::O_DIRECTORY|OFlag::O_NOFOLLOW, Mode::empty()) {
732 Ok(dir) => dir,
733 Err(nix::Error::Sys(Errno::ENOENT)) => {
734 self.report_vanished_file(&self.full_path())?;
735 continue; // fixme!!
736 },
737 Err(err) => bail!("open dir {:?} failed - {}", self.full_path(), err),
738 };
739
740 let child_magic = if dir_stat.st_dev != stat.st_dev {
741 detect_fs_type(dir.as_raw_fd())?
742 } else {
743 magic
744 };
745
746 self.write_filename(&filename)?;
747 self.encode_dir(&mut dir, &stat, child_magic, exclude_list)?;
748
749 } else if is_reg_file(&stat) {
750
751 let mut hardlink_target = None;
752
753 if stat.st_nlink > 1 {
754 let link_info = HardLinkInfo { st_dev: stat.st_dev, st_ino: stat.st_ino };
755 hardlink_target = self.hardlinks.get(&link_info).map(|(v, offset)| {
756 let mut target = v.clone().into_os_string();
757 target.push("\0"); // add Nul byte
758 (target, (start_pos as u64) - offset)
759 });
760 if hardlink_target == None {
761 self.hardlinks.insert(link_info, (self.relative_path.clone(), start_pos as u64));
762 }
763 }
764
765 if let Some((target, offset)) = hardlink_target {
766
767 self.write_filename(&filename)?;
768 self.encode_hardlink(target.as_bytes(), offset)?;
769
770 } else {
771
772 let filefd = match nix::fcntl::openat(rawfd, filename.as_ref(), OFlag::O_NOFOLLOW, Mode::empty()) {
773 Ok(filefd) => filefd,
774 Err(nix::Error::Sys(Errno::ENOENT)) => {
775 self.report_vanished_file(&self.full_path())?;
776 continue;
777 },
778 Err(err) => bail!("open file {:?} failed - {}", self.full_path(), err),
779 };
780
781 let child_magic = if dir_stat.st_dev != stat.st_dev {
782 detect_fs_type(filefd)?
783 } else {
784 magic
785 };
786
787 self.write_filename(&filename)?;
788 let res = self.encode_file(filefd, &stat, child_magic);
789 let _ = nix::unistd::close(filefd); // ignore close errors
790 res?;
791 }
792
793 } else if is_symlink(&stat) {
794 let mut buffer = vec::undefined(libc::PATH_MAX as usize);
795
796 let res = filename.with_nix_path(|cstr| {
797 unsafe { libc::readlinkat(rawfd, cstr.as_ptr(), buffer.as_mut_ptr() as *mut libc::c_char, buffer.len()-1) }
798 })?;
799
800 match Errno::result(res) {
801 Ok(len) => {
802 buffer[len as usize] = 0u8; // add Nul byte
803 self.write_filename(&filename)?;
804 self.encode_symlink(&buffer[..((len+1) as usize)], &stat)?
805 }
806 Err(nix::Error::Sys(Errno::ENOENT)) => {
807 self.report_vanished_file(&self.full_path())?;
808 continue;
809 }
810 Err(err) => bail!("readlink {:?} failed - {}", self.full_path(), err),
811 }
812 } else if is_block_dev(&stat) || is_char_dev(&stat) {
813 self.write_filename(&filename)?;
814 self.encode_device(&stat)?;
815 } else if is_fifo(&stat) || is_socket(&stat) {
816 self.write_filename(&filename)?;
817 self.encode_special(&stat)?;
818 } else {
819 bail!("unsupported file type (mode {:o} {:?})", stat.st_mode, self.full_path());
820 }
821
822 let end_pos = self.writer_pos;
823
824 goodbye_items.push(CaFormatGoodbyeItem {
825 offset: start_pos as u64,
826 size: (end_pos - start_pos) as u64,
827 hash: compute_goodbye_hash(filename.to_bytes()),
828 });
829
830 self.relative_path.pop();
831 }
832
833 //println!("encode_dir: {:?} end {}", self.full_path(), self.writer_pos);
834
835 // fixup goodby item offsets
836 let goodbye_start = self.writer_pos as u64;
837 for item in &mut goodbye_items {
838 item.offset = goodbye_start - item.offset;
839 }
840
841 let goodbye_offset = self.writer_pos - dir_start_pos;
842
843 self.write_goodbye_table(goodbye_offset, &mut goodbye_items)?;
844
845 //println!("encode_dir: {:?} end1 {}", self.full_path(), self.writer_pos);
846 Ok(())
847 }
848
849 fn encode_file(&mut self, filefd: RawFd, stat: &FileStat, magic: i64) -> Result<(), Error> {
850
851 //println!("encode_file: {:?}", self.full_path());
852
853 let mut entry = self.create_entry(&stat)?;
854
855 self.read_chattr(filefd, &mut entry)?;
856 self.read_fat_attr(filefd, magic, &mut entry)?;
857 let (xattrs, fcaps) = self.read_xattrs(filefd, &stat)?;
858 let acl_access = self.read_acl(filefd, &stat, acl::ACL_TYPE_ACCESS)?;
859 let projid = self.read_quota_project_id(filefd, magic, &stat)?;
860
861 self.write_entry(entry)?;
862 for xattr in xattrs {
863 self.write_xattr(xattr)?;
864 }
865 self.write_fcaps(fcaps)?;
866 for user in acl_access.users {
867 self.write_acl_user(user)?;
868 }
869 for group in acl_access.groups {
870 self.write_acl_group(group)?;
871 }
872 if let Some(group_obj) = acl_access.group_obj {
873 self.write_acl_group_obj(group_obj)?;
874 }
875 if let Some(projid) = projid {
876 self.write_quota_project_id(projid)?;
877 }
878
879 let include_payload;
880 if is_virtual_file_system(magic) {
881 include_payload = false;
882 } else {
883 if let Some(ref set) = &self.device_set {
884 include_payload = set.contains(&stat.st_dev);
885 } else {
886 include_payload = true;
887 }
888 }
889
890 if !include_payload {
891 eprintln!("skip content: {:?}", self.full_path());
892 self.write_header(CA_FORMAT_PAYLOAD, 0)?;
893 return Ok(());
894 }
895
896 let size = stat.st_size as u64;
897
898 self.write_header(CA_FORMAT_PAYLOAD, size)?;
899
900 let mut pos: u64 = 0;
901 loop {
902 let n = match nix::unistd::read(filefd, &mut self.file_copy_buffer) {
903 Ok(n) => n,
904 Err(nix::Error::Sys(Errno::EINTR)) => continue /* try again */,
905 Err(err) => bail!("read {:?} failed - {}", self.full_path(), err),
906 };
907 if n == 0 { // EOF
908 if pos != size {
909 // Note:: casync format cannot handle that
910 bail!("detected shrinked file {:?} ({} < {})", self.full_path(), pos, size);
911 }
912 break;
913 }
914
915 let mut next = pos + (n as u64);
916
917 if next > size { next = size; }
918
919 let count = (next - pos) as usize;
920
921 self.flush_copy_buffer(count)?;
922
923 pos = next;
924
925 if pos >= size { break; }
926 }
927
928 Ok(())
929 }
930
931 fn encode_device(&mut self, stat: &FileStat) -> Result<(), Error> {
932
933 let entry = self.create_entry(&stat)?;
934
935 self.write_entry(entry)?;
936
937 let major = unsafe { libc::major(stat.st_rdev) } as u64;
938 let minor = unsafe { libc::minor(stat.st_rdev) } as u64;
939
940 //println!("encode_device: {:?} {} {} {}", self.full_path(), stat.st_rdev, major, minor);
941
942 self.write_header(CA_FORMAT_DEVICE, std::mem::size_of::<CaFormatDevice>() as u64)?;
943 self.write_item(CaFormatDevice { major, minor })?;
944
945 Ok(())
946 }
947
948 // FIFO or Socket
949 fn encode_special(&mut self, stat: &FileStat) -> Result<(), Error> {
950
951 let entry = self.create_entry(&stat)?;
952
953 self.write_entry(entry)?;
954
955 Ok(())
956 }
957
958 fn encode_symlink(&mut self, target: &[u8], stat: &FileStat) -> Result<(), Error> {
959
960 //println!("encode_symlink: {:?} -> {:?}", self.full_path(), target);
961
962 let entry = self.create_entry(&stat)?;
963 self.write_entry(entry)?;
964
965 self.write_header(CA_FORMAT_SYMLINK, target.len() as u64)?;
966 self.write(target)?;
967
968 Ok(())
969 }
970
971 fn encode_hardlink(&mut self, target: &[u8], offset: u64) -> Result<(), Error> {
972
973 //println!("encode_hardlink: {:?} -> {:?}", self.full_path(), target);
974
975 // Note: HARDLINK replaces an ENTRY.
976 self.write_header(PXAR_FORMAT_HARDLINK, (target.len() as u64) + 8)?;
977 self.write_item(offset)?;
978 self.write(target)?;
979
980 Ok(())
981 }
982
983 fn encode_pxar_exclude(&mut self, filefd: RawFd, stat: &FileStat, magic: i64, content: &[u8]) -> Result<(), Error> {
984 let mut entry = self.create_entry(&stat)?;
985
986 self.read_chattr(filefd, &mut entry)?;
987 self.read_fat_attr(filefd, magic, &mut entry)?;
988 let (xattrs, fcaps) = self.read_xattrs(filefd, &stat)?;
989 let acl_access = self.read_acl(filefd, &stat, acl::ACL_TYPE_ACCESS)?;
990 let projid = self.read_quota_project_id(filefd, magic, &stat)?;
991
992 self.write_entry(entry)?;
993 for xattr in xattrs {
994 self.write_xattr(xattr)?;
995 }
996 self.write_fcaps(fcaps)?;
997 for user in acl_access.users {
998 self.write_acl_user(user)?;
999 }
1000 for group in acl_access.groups {
1001 self.write_acl_group(group)?;
1002 }
1003 if let Some(group_obj) = acl_access.group_obj {
1004 self.write_acl_group_obj(group_obj)?;
1005 }
1006 if let Some(projid) = projid {
1007 self.write_quota_project_id(projid)?;
1008 }
1009
1010 let include_payload;
1011 if is_virtual_file_system(magic) {
1012 include_payload = false;
1013 } else {
1014 if let Some(set) = &self.device_set {
1015 include_payload = set.contains(&stat.st_dev);
1016 } else {
1017 include_payload = true;
1018 }
1019 }
1020
1021 if !include_payload {
1022 eprintln!("skip content: {:?}", self.full_path());
1023 self.write_header(CA_FORMAT_PAYLOAD, 0)?;
1024 return Ok(());
1025 }
1026
1027 let size = content.len();
1028 self.write_header(CA_FORMAT_PAYLOAD, size as u64)?;
1029 self.writer.write_all(content)?;
1030 self.writer_pos += size;
1031
1032 Ok(())
1033 }
1034
1035 // the report_XXX method may raise and error - depending on encoder configuration
1036
1037 fn report_vanished_file(&self, path: &Path) -> Result<(), Error> {
1038
1039 eprintln!("WARNING: detected vanished file {:?}", path);
1040
1041 Ok(())
1042 }
1043 }
1044
1045 // If there is a match, an updated PxarExcludePattern list to pass to the matched child is returned.
1046 fn match_exclude_pattern(
1047 filename: &CStr,
1048 stat: &FileStat,
1049 match_pattern: &Vec<PxarExcludePattern>
1050 ) -> (MatchType, Vec<PxarExcludePattern>) {
1051 let mut child_pattern = Vec::new();
1052 let mut match_state = MatchType::None;
1053
1054 for pattern in match_pattern {
1055 match pattern.matches_filename(filename, is_directory(&stat)) {
1056 MatchType::None => {},
1057 MatchType::Exclude => match_state = MatchType::Exclude,
1058 MatchType::Include => match_state = MatchType::Include,
1059 MatchType::PartialExclude => {
1060 if match_state != MatchType::Exclude && match_state != MatchType::Include {
1061 match_state = MatchType::PartialExclude;
1062 }
1063 child_pattern.push(pattern.get_rest_pattern());
1064 },
1065 MatchType::PartialInclude => {
1066 if match_state != MatchType::Exclude && match_state != MatchType::Include {
1067 match_state = MatchType::PartialInclude;
1068 }
1069 child_pattern.push(pattern.get_rest_pattern());
1070 },
1071 }
1072 }
1073
1074 (match_state, child_pattern)
1075 }
1076
1077 fn errno_is_unsupported(errno: Errno) -> bool {
1078
1079 match errno {
1080 Errno::ENOTTY | Errno::ENOSYS | Errno::EBADF | Errno::EOPNOTSUPP | Errno::EINVAL => {
1081 true
1082 }
1083 _ => false,
1084 }
1085 }
1086
1087 fn detect_fs_type(fd: RawFd) -> Result<i64, Error> {
1088 let mut fs_stat: libc::statfs = unsafe { std::mem::uninitialized() };
1089 let res = unsafe { libc::fstatfs(fd, &mut fs_stat) };
1090 Errno::result(res)?;
1091
1092 Ok(fs_stat.f_type)
1093 }
1094
1095 #[inline(always)]
1096 pub fn is_temporary_file_system(magic: i64) -> bool {
1097 magic == RAMFS_MAGIC || magic == TMPFS_MAGIC
1098 }
1099
1100 pub fn is_virtual_file_system(magic: i64) -> bool {
1101
1102 match magic {
1103 BINFMTFS_MAGIC |
1104 CGROUP2_SUPER_MAGIC |
1105 CGROUP_SUPER_MAGIC |
1106 CONFIGFS_MAGIC |
1107 DEBUGFS_MAGIC |
1108 DEVPTS_SUPER_MAGIC |
1109 EFIVARFS_MAGIC |
1110 FUSE_CTL_SUPER_MAGIC |
1111 HUGETLBFS_MAGIC |
1112 MQUEUE_MAGIC |
1113 NFSD_MAGIC |
1114 PROC_SUPER_MAGIC |
1115 PSTOREFS_MAGIC |
1116 RPCAUTH_GSSMAGIC |
1117 SECURITYFS_MAGIC |
1118 SELINUX_MAGIC |
1119 SMACK_MAGIC |
1120 SYSFS_MAGIC => true,
1121 _ => false
1122 }
1123 }