src/backup/fixed_index.rs

   1 use anyhow::{bail, format_err, Error};
   2 use std::io::{Seek, SeekFrom};
   3
   4 use super::chunk_stat::*;
   5 use super::chunk_store::*;
   6 use super::{ChunkReadInfo, IndexFile};
   7 use crate::tools;
   8
   9 use std::fs::File;
  10 use std::io::Write;
  11 use std::os::unix::io::AsRawFd;
  12 use std::path::{Path, PathBuf};
  13 use std::sync::Arc;
  14
  15 use super::ChunkInfo;
  16
  17 use proxmox::tools::io::ReadExt;
  18 use proxmox::tools::Uuid;
  19
  20 /// Header format definition for fixed index files (`.fidx`)
  21 #[repr(C)]
  22 pub struct FixedIndexHeader {
  23     pub magic: [u8; 8],
  24     pub uuid: [u8; 16],
  25     pub ctime: i64,
  26     /// Sha256 over the index ``SHA256(digest1||digest2||...)``
  27     pub index_csum: [u8; 32],
  28     pub size: u64,
  29     pub chunk_size: u64,
  30     reserved: [u8; 4016], // overall size is one page (4096 bytes)
  31 }
  32 proxmox::static_assert_size!(FixedIndexHeader, 4096);
  33
  34 // split image into fixed size chunks
  35
  36 pub struct FixedIndexReader {
  37     _file: File,
  38     pub chunk_size: usize,
  39     pub size: u64,
  40     index_length: usize,
  41     index: *mut u8,
  42     pub uuid: [u8; 16],
  43     pub ctime: i64,
  44     pub index_csum: [u8; 32],
  45 }
  46
  47 // `index` is mmap()ed which cannot be thread-local so should be sendable
  48 unsafe impl Send for FixedIndexReader {}
  49 unsafe impl Sync for FixedIndexReader {}
  50
  51 impl Drop for FixedIndexReader {
  52     fn drop(&mut self) {
  53         if let Err(err) = self.unmap() {
  54             eprintln!("Unable to unmap file - {}", err);
  55         }
  56     }
  57 }
  58
  59 impl FixedIndexReader {
  60     pub fn open(path: &Path) -> Result<Self, Error> {
  61         File::open(path)
  62             .map_err(Error::from)
  63             .and_then(|file| Self::new(file))
  64             .map_err(|err| format_err!("Unable to open fixed index {:?} - {}", path, err))
  65     }
  66
  67     pub fn new(mut file: std::fs::File) -> Result<Self, Error> {
  68         file.seek(SeekFrom::Start(0))?;
  69
  70         let header_size = std::mem::size_of::<FixedIndexHeader>();
  71
  72         let stat = match nix::sys::stat::fstat(file.as_raw_fd()) {
  73             Ok(stat) => stat,
  74             Err(err) => bail!("fstat failed - {}", err),
  75         };
  76
  77         let size = stat.st_size as usize;
  78
  79         if size < header_size {
  80             bail!("index too small ({})", stat.st_size);
  81         }
  82
  83         let header: Box<FixedIndexHeader> = unsafe { file.read_host_value_boxed()? };
  84
  85         if header.magic != super::FIXED_SIZED_CHUNK_INDEX_1_0 {
  86             bail!("got unknown magic number");
  87         }
  88
  89         let size = u64::from_le(header.size);
  90         let ctime = i64::from_le(header.ctime);
  91         let chunk_size = u64::from_le(header.chunk_size);
  92
  93         let index_length = ((size + chunk_size - 1) / chunk_size) as usize;
  94         let index_size = index_length * 32;
  95
  96         let expected_index_size = (stat.st_size as usize) - header_size;
  97         if index_size != expected_index_size {
  98             bail!(
  99                 "got unexpected file size ({} != {})",
 100                 index_size,
 101                 expected_index_size
 102             );
 103         }
 104
 105         let data = unsafe {
 106             nix::sys::mman::mmap(
 107                 std::ptr::null_mut(),
 108                 index_size,
 109                 nix::sys::mman::ProtFlags::PROT_READ,
 110                 nix::sys::mman::MapFlags::MAP_PRIVATE,
 111                 file.as_raw_fd(),
 112                 header_size as i64,
 113             )
 114         }? as *mut u8;
 115
 116         Ok(Self {
 117             _file: file,
 118             chunk_size: chunk_size as usize,
 119             size,
 120             index_length,
 121             index: data,
 122             ctime,
 123             uuid: header.uuid,
 124             index_csum: header.index_csum,
 125         })
 126     }
 127
 128     fn unmap(&mut self) -> Result<(), Error> {
 129         if self.index == std::ptr::null_mut() {
 130             return Ok(());
 131         }
 132
 133         let index_size = self.index_length * 32;
 134
 135         if let Err(err) =
 136             unsafe { nix::sys::mman::munmap(self.index as *mut std::ffi::c_void, index_size) }
 137         {
 138             bail!("unmap file failed - {}", err);
 139         }
 140
 141         self.index = std::ptr::null_mut();
 142
 143         Ok(())
 144     }
 145
 146     pub fn print_info(&self) {
 147         println!("Size: {}", self.size);
 148         println!("ChunkSize: {}", self.chunk_size);
 149
 150         let mut ctime_str = self.ctime.to_string();
 151         if let Ok(s) = proxmox::tools::time::strftime_local("%c", self.ctime) {
 152             ctime_str = s;
 153         }
 154
 155         println!("CTime: {}", ctime_str);
 156         println!("UUID: {:?}", self.uuid);
 157     }
 158 }
 159
 160 impl IndexFile for FixedIndexReader {
 161     fn index_count(&self) -> usize {
 162         self.index_length
 163     }
 164
 165     fn index_digest(&self, pos: usize) -> Option<&[u8; 32]> {
 166         if pos >= self.index_length {
 167             None
 168         } else {
 169             Some(unsafe { std::mem::transmute(self.index.add(pos * 32)) })
 170         }
 171     }
 172
 173     fn index_bytes(&self) -> u64 {
 174         self.size
 175     }
 176
 177     fn chunk_info(&self, pos: usize) -> Option<ChunkReadInfo> {
 178         if pos >= self.index_length {
 179             return None;
 180         }
 181
 182         let start = (pos * self.chunk_size) as u64;
 183         let mut end = start + self.chunk_size as u64;
 184
 185         if end > self.size {
 186             end = self.size;
 187         }
 188
 189         let digest = self.index_digest(pos).unwrap();
 190         Some(ChunkReadInfo {
 191             range: start..end,
 192             digest: *digest,
 193         })
 194     }
 195
 196     fn compute_csum(&self) -> ([u8; 32], u64) {
 197         let mut csum = openssl::sha::Sha256::new();
 198         let mut chunk_end = 0;
 199         for pos in 0..self.index_count() {
 200             let info = self.chunk_info(pos).unwrap();
 201             chunk_end = info.range.end;
 202             csum.update(&info.digest);
 203         }
 204         let csum = csum.finish();
 205
 206         (csum, chunk_end)
 207     }
 208
 209     fn chunk_from_offset(&self, offset: u64) -> Option<(usize, u64)> {
 210         if offset >= self.size {
 211             return None;
 212         }
 213
 214         Some((
 215             (offset / self.chunk_size as u64) as usize,
 216             offset & (self.chunk_size - 1) as u64, // fast modulo, valid for 2^x chunk_size
 217         ))
 218     }
 219 }
 220
 221 pub struct FixedIndexWriter {
 222     store: Arc<ChunkStore>,
 223     file: File,
 224     _lock: tools::ProcessLockSharedGuard,
 225     filename: PathBuf,
 226     tmp_filename: PathBuf,
 227     chunk_size: usize,
 228     size: usize,
 229     index_length: usize,
 230     index: *mut u8,
 231     pub uuid: [u8; 16],
 232     pub ctime: i64,
 233 }
 234
 235 // `index` is mmap()ed which cannot be thread-local so should be sendable
 236 unsafe impl Send for FixedIndexWriter {}
 237
 238 impl Drop for FixedIndexWriter {
 239     fn drop(&mut self) {
 240         let _ = std::fs::remove_file(&self.tmp_filename); // ignore errors
 241         if let Err(err) = self.unmap() {
 242             eprintln!("Unable to unmap file {:?} - {}", self.tmp_filename, err);
 243         }
 244     }
 245 }
 246
 247 impl FixedIndexWriter {
 248     #[allow(clippy::cast_ptr_alignment)]
 249     pub fn create(
 250         store: Arc<ChunkStore>,
 251         path: &Path,
 252         size: usize,
 253         chunk_size: usize,
 254     ) -> Result<Self, Error> {
 255         let shared_lock = store.try_shared_lock()?;
 256
 257         let full_path = store.relative_path(path);
 258         let mut tmp_path = full_path.clone();
 259         tmp_path.set_extension("tmp_fidx");
 260
 261         let mut file = std::fs::OpenOptions::new()
 262             .create(true)
 263             .truncate(true)
 264             .read(true)
 265             .write(true)
 266             .open(&tmp_path)?;
 267
 268         let header_size = std::mem::size_of::<FixedIndexHeader>();
 269
 270         // todo: use static assertion when available in rust
 271         if header_size != 4096 {
 272             panic!("got unexpected header size");
 273         }
 274
 275         let ctime = proxmox::tools::time::epoch_i64();
 276
 277         let uuid = Uuid::generate();
 278
 279         let buffer = vec![0u8; header_size];
 280         let header = unsafe { &mut *(buffer.as_ptr() as *mut FixedIndexHeader) };
 281
 282         header.magic = super::FIXED_SIZED_CHUNK_INDEX_1_0;
 283         header.ctime = i64::to_le(ctime);
 284         header.size = u64::to_le(size as u64);
 285         header.chunk_size = u64::to_le(chunk_size as u64);
 286         header.uuid = *uuid.as_bytes();
 287
 288         header.index_csum = [0u8; 32];
 289
 290         file.write_all(&buffer)?;
 291
 292         let index_length = (size + chunk_size - 1) / chunk_size;
 293         let index_size = index_length * 32;
 294         nix::unistd::ftruncate(file.as_raw_fd(), (header_size + index_size) as i64)?;
 295
 296         let data = unsafe {
 297             nix::sys::mman::mmap(
 298                 std::ptr::null_mut(),
 299                 index_size,
 300                 nix::sys::mman::ProtFlags::PROT_READ | nix::sys::mman::ProtFlags::PROT_WRITE,
 301                 nix::sys::mman::MapFlags::MAP_SHARED,
 302                 file.as_raw_fd(),
 303                 header_size as i64,
 304             )
 305         }? as *mut u8;
 306
 307         Ok(Self {
 308             store,
 309             file,
 310             _lock: shared_lock,
 311             filename: full_path,
 312             tmp_filename: tmp_path,
 313             chunk_size,
 314             size,
 315             index_length,
 316             index: data,
 317             ctime,
 318             uuid: *uuid.as_bytes(),
 319         })
 320     }
 321
 322     pub fn index_length(&self) -> usize {
 323         self.index_length
 324     }
 325
 326     fn unmap(&mut self) -> Result<(), Error> {
 327         if self.index == std::ptr::null_mut() {
 328             return Ok(());
 329         }
 330
 331         let index_size = self.index_length * 32;
 332
 333         if let Err(err) =
 334             unsafe { nix::sys::mman::munmap(self.index as *mut std::ffi::c_void, index_size) }
 335         {
 336             bail!("unmap file {:?} failed - {}", self.tmp_filename, err);
 337         }
 338
 339         self.index = std::ptr::null_mut();
 340
 341         Ok(())
 342     }
 343
 344     pub fn close(&mut self) -> Result<[u8; 32], Error> {
 345         if self.index == std::ptr::null_mut() {
 346             bail!("cannot close already closed index file.");
 347         }
 348
 349         let index_size = self.index_length * 32;
 350         let data = unsafe { std::slice::from_raw_parts(self.index, index_size) };
 351         let index_csum = openssl::sha::sha256(data);
 352
 353         self.unmap()?;
 354
 355         let csum_offset = proxmox::offsetof!(FixedIndexHeader, index_csum);
 356         self.file.seek(SeekFrom::Start(csum_offset as u64))?;
 357         self.file.write_all(&index_csum)?;
 358         self.file.flush()?;
 359
 360         if let Err(err) = std::fs::rename(&self.tmp_filename, &self.filename) {
 361             bail!("Atomic rename file {:?} failed - {}", self.filename, err);
 362         }
 363
 364         Ok(index_csum)
 365     }
 366
 367     pub fn check_chunk_alignment(&self, offset: usize, chunk_len: usize) -> Result<usize, Error> {
 368         if offset < chunk_len {
 369             bail!("got chunk with small offset ({} < {}", offset, chunk_len);
 370         }
 371
 372         let pos = offset - chunk_len;
 373
 374         if offset > self.size {
 375             bail!("chunk data exceeds size ({} >= {})", offset, self.size);
 376         }
 377
 378         // last chunk can be smaller
 379         if ((offset != self.size) && (chunk_len != self.chunk_size))
 380             || (chunk_len > self.chunk_size)
 381             || (chunk_len == 0)
 382         {
 383             bail!(
 384                 "chunk with unexpected length ({} != {}",
 385                 chunk_len,
 386                 self.chunk_size
 387             );
 388         }
 389
 390         if pos & (self.chunk_size - 1) != 0 {
 391             bail!("got unaligned chunk (pos = {})", pos);
 392         }
 393
 394         Ok(pos / self.chunk_size)
 395     }
 396
 397     // Note: We want to add data out of order, so do not assume any order here.
 398     pub fn add_chunk(&mut self, chunk_info: &ChunkInfo, stat: &mut ChunkStat) -> Result<(), Error> {
 399         let chunk_len = chunk_info.chunk_len as usize;
 400         let offset = chunk_info.offset as usize; // end of chunk
 401
 402         let idx = self.check_chunk_alignment(offset, chunk_len)?;
 403
 404         let (is_duplicate, compressed_size) = self
 405             .store
 406             .insert_chunk(&chunk_info.chunk, &chunk_info.digest)?;
 407
 408         stat.chunk_count += 1;
 409         stat.compressed_size += compressed_size;
 410
 411         let digest = &chunk_info.digest;
 412
 413         println!(
 414             "ADD CHUNK {} {} {}% {} {}",
 415             idx,
 416             chunk_len,
 417             (compressed_size * 100) / (chunk_len as u64),
 418             is_duplicate,
 419             proxmox::tools::digest_to_hex(digest)
 420         );
 421
 422         if is_duplicate {
 423             stat.duplicate_chunks += 1;
 424         } else {
 425             stat.disk_size += compressed_size;
 426         }
 427
 428         self.add_digest(idx, digest)
 429     }
 430
 431     pub fn add_digest(&mut self, index: usize, digest: &[u8; 32]) -> Result<(), Error> {
 432         if index >= self.index_length {
 433             bail!(
 434                 "add digest failed - index out of range ({} >= {})",
 435                 index,
 436                 self.index_length
 437             );
 438         }
 439
 440         if self.index == std::ptr::null_mut() {
 441             bail!("cannot write to closed index file.");
 442         }
 443
 444         let index_pos = index * 32;
 445         unsafe {
 446             let dst = self.index.add(index_pos);
 447             dst.copy_from_nonoverlapping(digest.as_ptr(), 32);
 448         }
 449
 450         Ok(())
 451     }
 452
 453     pub fn clone_data_from(&mut self, reader: &FixedIndexReader) -> Result<(), Error> {
 454         if self.index_length != reader.index_count() {
 455             bail!("clone_data_from failed - index sizes not equal");
 456         }
 457
 458         for i in 0..self.index_length {
 459             self.add_digest(i, reader.index_digest(i).unwrap())?;
 460         }
 461
 462         Ok(())
 463     }
 464 }