1 //! *pxar* format encoder.
3 //! This module contain the code to generate *pxar* archive files.
6 use endian_trait
::Endian
;
7 use std
::collections
::HashMap
;
9 use super::format_definition
::*;
10 use super::binary_search_tree
::*;
11 use crate::tools
::xattr
;
14 use std
::os
::unix
::io
::AsRawFd
;
15 use std
::os
::unix
::ffi
::OsStrExt
;
16 use std
::os
::unix
::io
::RawFd
;
17 use std
::path
::{Path, PathBuf}
;
22 use nix
::fcntl
::OFlag
;
23 use nix
::sys
::stat
::Mode
;
24 use nix
::errno
::Errno
;
25 use nix
::sys
::stat
::FileStat
;
27 /// The format requires to build sorted directory lookup tables in
28 /// memory, so we restrict the number of allowed entries to limit
29 /// maximum memory usage.
30 pub const MAX_DIRECTORY_ENTRIES
: usize = 256*1024;
32 #[derive(Eq, PartialEq, Hash)]
38 pub struct Encoder
<'a
, W
: Write
> {
40 relative_path
: PathBuf
,
44 file_copy_buffer
: Vec
<u8>,
45 all_file_systems
: bool
,
48 hardlinks
: HashMap
<HardLinkInfo
, (PathBuf
, u64)>,
51 impl <'a
, W
: Write
> Encoder
<'a
, W
> {
53 // used for error reporting
54 fn full_path(&self) -> PathBuf
{
55 self.base_path
.join(&self.relative_path
)
60 dir
: &mut nix
::dir
::Dir
,
62 all_file_systems
: bool
,
64 ) -> Result
<(), Error
> {
66 const FILE_COPY_BUFFER_SIZE
: usize = 1024*1024;
68 let mut file_copy_buffer
= Vec
::with_capacity(FILE_COPY_BUFFER_SIZE
);
69 unsafe { file_copy_buffer.set_len(FILE_COPY_BUFFER_SIZE); }
72 // todo: use scandirat??
74 let dir_fd
= dir
.as_raw_fd();
75 let stat
= match nix
::sys
::stat
::fstat(dir_fd
) {
77 Err(err
) => bail
!("fstat {:?} failed - {}", path
, err
),
80 if (stat
.st_mode
& libc
::S_IFMT
) != libc
::S_IFDIR
{
81 bail
!("got unexpected file type {:?} (not a directory)", path
);
84 let magic
= detect_fs_type(dir_fd
)?
;
86 if is_virtual_file_system(magic
) {
87 bail
!("backup virtual file systems is disabled!");
92 relative_path
: PathBuf
::new(),
98 root_st_dev
: stat
.st_dev
,
100 hardlinks
: HashMap
::new(),
103 if verbose { println!("{:?}
", me.full_path()); }
105 me.encode_dir(dir, &stat, magic)?;
110 fn write(&mut self, buf: &[u8]) -> Result<(), Error> {
111 self.writer.write_all(buf)?;
112 self.writer_pos += buf.len();
116 fn write_item<T: Endian>(&mut self, item: T) -> Result<(), Error> {
118 let data = item.to_le();
120 let buffer = unsafe { std::slice::from_raw_parts(
121 &data as *const T as *const u8,
122 std::mem::size_of::<T>()
130 fn flush_copy_buffer(&mut self, size: usize) -> Result<(), Error> {
131 self.writer.write_all(&self.file_copy_buffer[..size])?;
132 self.writer_pos += size;
136 fn write_header(&mut self, htype: u64, size: u64) -> Result<(), Error> {
138 let size = size + (std::mem::size_of::<CaFormatHeader>() as u64);
139 self.write_item(CaFormatHeader { size, htype })?;
144 fn write_filename(&mut self, name: &CStr) -> Result<(), Error> {
146 let buffer = name.to_bytes_with_nul();
147 self.write_header(CA_FORMAT_FILENAME, buffer.len() as u64)?;
153 fn create_entry(&self, stat: &FileStat) -> Result<CaFormatEntry, Error> {
155 let mode = if (stat.st_mode & libc::S_IFMT) == libc::S_IFLNK {
156 (libc::S_IFLNK | 0o777) as u64
158 (stat.st_mode & (libc::S_IFMT | 0o7777)) as u64
161 let mtime = stat.st_mtime * 1_000_000_000 + stat.st_mtime_nsec;
163 bail!("got strange
mtime ({}
) from fstat
for {:?}
.", mtime, self.full_path());
167 let entry = CaFormatEntry {
168 feature_flags: CA_FORMAT_DEFAULT, // fixme: ??
171 uid: stat.st_uid as u64,
172 gid: stat.st_gid as u64,
179 fn read_chattr(&self, fd: RawFd, entry: &mut CaFormatEntry) -> Result<(), Error> {
181 let mut attr: usize = 0;
183 let res = unsafe { read_attr_fd(fd, &mut attr)};
184 if let Err(err) = res {
185 if let nix::Error::Sys(errno) = err {
186 if errno_is_unsupported(errno) { return Ok(()) };
188 bail!("read_attr_fd failed
for {:?}
- {}
", self.full_path(), err);
191 let flags = ca_feature_flags_from_chattr(attr as u32);
192 entry.flags = entry.flags | flags;
197 fn read_fat_attr(&self, fd: RawFd, magic: i64, entry: &mut CaFormatEntry) -> Result<(), Error> {
199 if magic != MSDOS_SUPER_MAGIC && magic != FUSE_SUPER_MAGIC { return Ok(()); }
201 let mut attr: u32 = 0;
203 let res = unsafe { read_fat_attr_fd(fd, &mut attr)};
204 if let Err(err) = res {
205 if let nix::Error::Sys(errno) = err {
206 if errno_is_unsupported(errno) { return Ok(()) };
208 bail!("read_fat_attr_fd failed
for {:?}
- {}
", self.full_path(), err);
211 let flags = ca_feature_flags_from_fat_attr(attr);
212 entry.flags = entry.flags | flags;
217 fn read_xattrs(&self, fd: RawFd, stat: &FileStat, entry: &CaFormatEntry) -> Result<(Vec<CaFormatXAttr>, Option<CaFormatFCaps>), Error> {
218 let mut xattrs = Vec::new();
219 let mut fcaps = None;
221 let flags = CA_FORMAT_WITH_XATTRS | CA_FORMAT_WITH_FCAPS;
222 if (entry.feature_flags & flags) == 0 { return Ok((xattrs, fcaps)); }
223 // Should never be called on symlinks, just in case check anyway
224 if (stat.st_mode & libc::S_IFMT) == libc::S_IFLNK { return Ok((xattrs, fcaps)); }
226 let xattr_names = match xattr::flistxattr(fd) {
228 Err(Errno::EOPNOTSUPP) => return Ok((xattrs, fcaps)),
229 Err(Errno::EBADF) => return Ok((xattrs, fcaps)),
230 Err(err) => bail!("read_xattrs failed
for {:?}
- {}
", self.full_path(), err),
233 for name in xattr_names.split(|c| *c == '\0' as u8) {
234 // Only extract the relevant extended attributes
235 if !xattr::name_store(&name) { continue; }
237 let value = match xattr::fgetxattr(fd, name) {
239 // Vanished between flistattr and getxattr, this is ok, silently ignore
240 Err(Errno::ENODATA) => continue,
241 Err(err) => bail!("read_xattrs failed
for {:?}
- {}
", self.full_path(), err),
244 if xattr::security_capability(&name) {
245 // fcaps are stored in own format within the archive
246 fcaps = Some(CaFormatFCaps {
250 xattrs.push(CaFormatXAttr {
261 fn write_entry(&mut self, entry: CaFormatEntry) -> Result<(), Error> {
263 self.write_header(CA_FORMAT_ENTRY, std::mem::size_of::<CaFormatEntry>() as u64)?;
264 self.write_item(entry)?;
269 fn write_xattr(&mut self, xattr: CaFormatXAttr) -> Result<(), Error> {
270 let size = xattr.name.len() + xattr.value.len() + 1; // +1 for '\0' separating name and value
271 self.write_header(CA_FORMAT_XATTR, size as u64)?;
272 self.write(xattr.name.as_slice())?;
274 self.write(xattr.value.as_slice())?;
279 fn write_fcaps(&mut self, fcaps: Option<CaFormatFCaps>) -> Result<(), Error> {
280 if let Some(fcaps) = fcaps {
281 let size = fcaps.data.len();
282 self.write_header(CA_FORMAT_FCAPS, size as u64)?;
283 self.write(fcaps.data.as_slice())?;
289 fn write_goodbye_table(&mut self, goodbye_offset: usize, goodbye_items: &mut [CaFormatGoodbyeItem]) -> Result<(), Error> {
291 goodbye_items.sort_unstable_by(|a, b| a.hash.cmp(&b.hash));
293 let item_count = goodbye_items.len();
295 let goodbye_table_size = (item_count + 1)*std::mem::size_of::<CaFormatGoodbyeItem>();
297 self.write_header(CA_FORMAT_GOODBYE, goodbye_table_size as u64)?;
299 if self.file_copy_buffer.len() < goodbye_table_size {
300 let need = goodbye_table_size - self.file_copy_buffer.len();
301 self.file_copy_buffer.reserve(need);
302 unsafe { self.file_copy_buffer.set_len(self.file_copy_buffer.capacity()); }
305 let buffer = &mut self.file_copy_buffer;
307 copy_binary_search_tree(item_count, |s, d| {
308 let item = &goodbye_items[s];
309 let offset = d*std::mem::size_of::<CaFormatGoodbyeItem>();
310 let dest = crate::tools::map_struct_mut::<CaFormatGoodbyeItem>(&mut buffer[offset..]).unwrap();
311 dest.offset = u64::to_le(item.offset);
312 dest.size = u64::to_le(item.size);
313 dest.hash = u64::to_le(item.hash);
316 // append CaFormatGoodbyeTail as last item
317 let offset = item_count*std::mem::size_of::<CaFormatGoodbyeItem>();
318 let dest = crate::tools::map_struct_mut::<CaFormatGoodbyeItem>(&mut buffer[offset..]).unwrap();
319 dest.offset = u64::to_le(goodbye_offset as u64);
320 dest.size = u64::to_le((goodbye_table_size + std::mem::size_of::<CaFormatHeader>()) as u64);
321 dest.hash = u64::to_le(CA_FORMAT_GOODBYE_TAIL_MARKER);
323 self.flush_copy_buffer(goodbye_table_size)?;
328 fn encode_dir(&mut self, dir: &mut nix::dir::Dir, dir_stat: &FileStat, magic: i64) -> Result<(), Error> {
330 //println!("encode_dir
: {:?} start {}
", self.full_path(), self.writer_pos);
332 let mut name_list = vec![];
334 let rawfd = dir.as_raw_fd();
336 let dir_start_pos = self.writer_pos;
338 let mut dir_entry = self.create_entry(&dir_stat)?;
340 self.read_chattr(rawfd, &mut dir_entry)?;
341 self.read_fat_attr(rawfd, magic, &mut dir_entry)?;
342 let (xattrs, fcaps) = self.read_xattrs(rawfd, &dir_stat, &dir_entry)?;
344 self.write_entry(dir_entry)?;
345 for xattr in xattrs { self.write_xattr(xattr)?; }
346 self.write_fcaps(fcaps)?;
348 let mut dir_count = 0;
350 let include_children;
351 if is_virtual_file_system(magic) {
352 include_children = false;
354 include_children = (self.root_st_dev == dir_stat.st_dev) || self.all_file_systems;
357 if include_children {
358 for entry in dir.iter() {
360 if dir_count > MAX_DIRECTORY_ENTRIES {
361 bail!("too many directory items
in {:?}
(> {}
)",
362 self.full_path(), MAX_DIRECTORY_ENTRIES);
365 let entry = match entry {
367 Err(err) => bail!("readir {:?} failed
- {}
", self.full_path(), err),
369 let filename = entry.file_name().to_owned();
371 let name = filename.to_bytes_with_nul();
372 let name_len = name.len();
373 if name_len == 2 && name[0] == b'.' && name[1] == 0u8 { continue; }
374 if name_len == 3 && name[0] == b'.' && name[1] == b'.' && name[2] == 0u8 { continue; }
376 name_list.push(filename);
379 eprintln!("skip mount point
: {:?}
", self.full_path());
382 name_list.sort_unstable_by(|a, b| a.cmp(&b));
384 let mut goodbye_items = vec![];
386 for filename in &name_list {
387 self.relative_path.push(std::ffi::OsStr::from_bytes(filename.as_bytes()));
389 if self.verbose { println!("{:?}", self.full_path()); }
391 let stat
= match nix
::sys
::stat
::fstatat(rawfd
, filename
.as_ref(), nix
::fcntl
::AtFlags
::AT_SYMLINK_NOFOLLOW
) {
393 Err(nix
::Error
::Sys(Errno
::ENOENT
)) => {
394 self.report_vanished_file(&self.full_path())?
;
397 Err(err
) => bail
!("fstat {:?} failed - {}", self.full_path(), err
),
400 let start_pos
= self.writer_pos
;
402 let ifmt
= stat
.st_mode
& libc
::S_IFMT
;
404 if ifmt
== libc
::S_IFDIR
{
406 let mut dir
= match nix
::dir
::Dir
::openat(rawfd
, filename
.as_ref(), OFlag
::O_DIRECTORY
|OFlag
::O_NOFOLLOW
, Mode
::empty()) {
408 Err(nix
::Error
::Sys(Errno
::ENOENT
)) => {
409 self.report_vanished_file(&self.full_path())?
;
412 Err(err
) => bail
!("open dir {:?} failed - {}", self.full_path(), err
),
415 let child_magic
= if dir_stat
.st_dev
!= stat
.st_dev
{
416 detect_fs_type(dir
.as_raw_fd())?
421 self.write_filename(&filename
)?
;
422 self.encode_dir(&mut dir
, &stat
, child_magic
)?
;
424 } else if ifmt
== libc
::S_IFREG
{
426 let mut hardlink_target
= None
;
428 if stat
.st_nlink
> 1 {
429 let link_info
= HardLinkInfo { st_dev: stat.st_dev, st_ino: stat.st_ino }
;
430 hardlink_target
= self.hardlinks
.get(&link_info
).map(|(v
, offset
)| {
431 let mut target
= v
.clone().into_os_string();
432 target
.push("\0"); // add Nul byte
433 (target
, (start_pos
as u64) - offset
)
435 if hardlink_target
== None
{
436 self.hardlinks
.insert(link_info
, (self.relative_path
.clone(), start_pos
as u64));
440 if let Some((target
, offset
)) = hardlink_target
{
442 self.write_filename(&filename
)?
;
443 self.encode_hardlink(target
.as_bytes(), offset
)?
;
447 let filefd
= match nix
::fcntl
::openat(rawfd
, filename
.as_ref(), OFlag
::O_NOFOLLOW
, Mode
::empty()) {
448 Ok(filefd
) => filefd
,
449 Err(nix
::Error
::Sys(Errno
::ENOENT
)) => {
450 self.report_vanished_file(&self.full_path())?
;
453 Err(err
) => bail
!("open file {:?} failed - {}", self.full_path(), err
),
456 let child_magic
= if dir_stat
.st_dev
!= stat
.st_dev
{
457 detect_fs_type(filefd
)?
462 self.write_filename(&filename
)?
;
463 let res
= self.encode_file(filefd
, &stat
, child_magic
);
464 let _
= nix
::unistd
::close(filefd
); // ignore close errors
468 } else if ifmt
== libc
::S_IFLNK
{
469 let mut buffer
= [0u8; libc
::PATH_MAX
as usize];
471 let res
= filename
.with_nix_path(|cstr
| {
472 unsafe { libc::readlinkat(rawfd, cstr.as_ptr(), buffer.as_mut_ptr() as *mut libc::c_char, buffer.len()-1) }
475 match Errno
::result(res
) {
477 buffer
[len
as usize] = 0u8; // add Nul byte
478 self.write_filename(&filename
)?
;
479 self.encode_symlink(&buffer
[..((len
+1) as usize)], &stat
)?
481 Err(nix
::Error
::Sys(Errno
::ENOENT
)) => {
482 self.report_vanished_file(&self.full_path())?
;
485 Err(err
) => bail
!("readlink {:?} failed - {}", self.full_path(), err
),
487 } else if (ifmt
== libc
::S_IFBLK
) || (ifmt
== libc
::S_IFCHR
) {
488 self.write_filename(&filename
)?
;
489 self.encode_device(&stat
)?
;
490 } else if (ifmt
== libc
::S_IFIFO
) || (ifmt
== libc
::S_IFSOCK
) {
491 self.write_filename(&filename
)?
;
492 self.encode_special(&stat
)?
;
494 bail
!("unsupported file type (mode {:o} {:?})", stat
.st_mode
, self.full_path());
497 let end_pos
= self.writer_pos
;
499 goodbye_items
.push(CaFormatGoodbyeItem
{
500 offset
: start_pos
as u64,
501 size
: (end_pos
- start_pos
) as u64,
502 hash
: compute_goodbye_hash(filename
.to_bytes()),
505 self.relative_path
.pop();
508 //println!("encode_dir: {:?} end {}", self.full_path(), self.writer_pos);
510 // fixup goodby item offsets
511 let goodbye_start
= self.writer_pos
as u64;
512 for item
in &mut goodbye_items
{
513 item
.offset
= goodbye_start
- item
.offset
;
516 let goodbye_offset
= self.writer_pos
- dir_start_pos
;
518 self.write_goodbye_table(goodbye_offset
, &mut goodbye_items
)?
;
520 //println!("encode_dir: {:?} end1 {}", self.full_path(), self.writer_pos);
524 fn encode_file(&mut self, filefd
: RawFd
, stat
: &FileStat
, magic
: i64) -> Result
<(), Error
> {
526 //println!("encode_file: {:?}", self.full_path());
528 let mut entry
= self.create_entry(&stat
)?
;
530 self.read_chattr(filefd
, &mut entry
)?
;
531 self.read_fat_attr(filefd
, magic
, &mut entry
)?
;
532 let (xattrs
, fcaps
) = self.read_xattrs(filefd
, &stat
, &entry
)?
;
534 self.write_entry(entry
)?
;
535 for xattr
in xattrs { self.write_xattr(xattr)?; }
536 self.write_fcaps(fcaps
)?
;
539 if is_virtual_file_system(magic
) {
540 include_payload
= false;
542 include_payload
= (stat
.st_dev
== self.root_st_dev
) || self.all_file_systems
;
545 if !include_payload
{
546 eprintln
!("skip content: {:?}", self.full_path());
547 self.write_header(CA_FORMAT_PAYLOAD
, 0)?
;
551 let size
= stat
.st_size
as u64;
553 self.write_header(CA_FORMAT_PAYLOAD
, size
)?
;
555 let mut pos
: u64 = 0;
557 let n
= match nix
::unistd
::read(filefd
, &mut self.file_copy_buffer
) {
559 Err(nix
::Error
::Sys(Errno
::EINTR
)) => continue /* try again */,
560 Err(err
) => bail
!("read {:?} failed - {}", self.full_path(), err
),
564 // Note:: casync format cannot handle that
565 bail
!("detected shrinked file {:?} ({} < {})", self.full_path(), pos
, size
);
570 let mut next
= pos
+ (n
as u64);
572 if next
> size { next = size; }
574 let count
= (next
- pos
) as usize;
576 self.flush_copy_buffer(count
)?
;
580 if pos
>= size { break; }
586 fn encode_device(&mut self, stat
: &FileStat
) -> Result
<(), Error
> {
588 let entry
= self.create_entry(&stat
)?
;
590 self.write_entry(entry
)?
;
592 let major
= unsafe { libc::major(stat.st_rdev) }
as u64;
593 let minor
= unsafe { libc::minor(stat.st_rdev) }
as u64;
595 //println!("encode_device: {:?} {} {} {}", self.full_path(), stat.st_rdev, major, minor);
597 self.write_header(CA_FORMAT_DEVICE
, std
::mem
::size_of
::<CaFormatDevice
>() as u64)?
;
598 self.write_item(CaFormatDevice { major, minor }
)?
;
604 fn encode_special(&mut self, stat
: &FileStat
) -> Result
<(), Error
> {
606 let entry
= self.create_entry(&stat
)?
;
608 self.write_entry(entry
)?
;
613 fn encode_symlink(&mut self, target
: &[u8], stat
: &FileStat
) -> Result
<(), Error
> {
615 //println!("encode_symlink: {:?} -> {:?}", self.full_path(), target);
617 let entry
= self.create_entry(&stat
)?
;
618 self.write_entry(entry
)?
;
620 self.write_header(CA_FORMAT_SYMLINK
, target
.len() as u64)?
;
626 fn encode_hardlink(&mut self, target
: &[u8], offset
: u64) -> Result
<(), Error
> {
628 //println!("encode_hardlink: {:?} -> {:?}", self.full_path(), target);
630 // Note: HARDLINK replaces an ENTRY.
631 self.write_header(PXAR_FORMAT_HARDLINK
, (target
.len() as u64) + 8)?
;
632 self.write_item(offset
)?
;
638 // the report_XXX method may raise and error - depending on encoder configuration
640 fn report_vanished_file(&self, path
: &Path
) -> Result
<(), Error
> {
642 eprintln
!("WARNING: detected vanished file {:?}", path
);
648 fn errno_is_unsupported(errno
: Errno
) -> bool
{
651 Errno
::ENOTTY
| Errno
::ENOSYS
| Errno
::EBADF
| Errno
::EOPNOTSUPP
| Errno
::EINVAL
=> {
658 fn detect_fs_type(fd
: RawFd
) -> Result
<i64, Error
> {
659 let mut fs_stat
: libc
::statfs
= unsafe { std::mem::uninitialized() }
;
660 let res
= unsafe { libc::fstatfs(fd, &mut fs_stat) }
;
666 use nix
::{convert_ioctl_res, request_code_read, ioc}
;
668 // /usr/include/linux/fs.h: #define FS_IOC_GETFLAGS _IOR('f', 1, long)
669 /// read Linux file system attributes (see man chattr)
670 nix
::ioctl_read
!(read_attr_fd
, b'f'
, 1, usize);
672 // /usr/include/linux/msdos_fs.h: #define FAT_IOCTL_GET_ATTRIBUTES _IOR('r', 0x10, __u32)
673 // read FAT file system attributes
674 nix
::ioctl_read
!(read_fat_attr_fd
, b'r'
, 0x10, u32);
677 // from /usr/include/linux/magic.h
678 // and from casync util.h
679 pub const BINFMTFS_MAGIC
: i64 = 0x42494e4d;
680 pub const CGROUP2_SUPER_MAGIC
: i64 = 0x63677270;
681 pub const CGROUP_SUPER_MAGIC
: i64 = 0x0027e0eb;
682 pub const CONFIGFS_MAGIC
: i64 = 0x62656570;
683 pub const DEBUGFS_MAGIC
: i64 = 0x64626720;
684 pub const DEVPTS_SUPER_MAGIC
: i64 = 0x00001cd1;
685 pub const EFIVARFS_MAGIC
: i64 = 0xde5e81e4;
686 pub const FUSE_CTL_SUPER_MAGIC
: i64 = 0x65735543;
687 pub const HUGETLBFS_MAGIC
: i64 = 0x958458f6;
688 pub const MQUEUE_MAGIC
: i64 = 0x19800202;
689 pub const NFSD_MAGIC
: i64 = 0x6e667364;
690 pub const PROC_SUPER_MAGIC
: i64 = 0x00009fa0;
691 pub const PSTOREFS_MAGIC
: i64 = 0x6165676C;
692 pub const RPCAUTH_GSSMAGIC
: i64 = 0x67596969;
693 pub const SECURITYFS_MAGIC
: i64 = 0x73636673;
694 pub const SELINUX_MAGIC
: i64 = 0xf97cff8c;
695 pub const SMACK_MAGIC
: i64 = 0x43415d53;
696 pub const RAMFS_MAGIC
: i64 = 0x858458f6;
697 pub const TMPFS_MAGIC
: i64 = 0x01021994;
698 pub const SYSFS_MAGIC
: i64 = 0x62656572;
699 pub const MSDOS_SUPER_MAGIC
: i64 = 0x00004d44;
700 pub const FUSE_SUPER_MAGIC
: i64 = 0x65735546;
704 pub fn is_temporary_file_system(magic
: i64) -> bool
{
705 magic
== RAMFS_MAGIC
|| magic
== TMPFS_MAGIC
708 pub fn is_virtual_file_system(magic
: i64) -> bool
{
712 CGROUP2_SUPER_MAGIC
|
718 FUSE_CTL_SUPER_MAGIC
|