src/backup/chunk_store.rs

   1 use failure::*;
   2 use std::path::{Path, PathBuf};
   3 use std::io::Write;
   4
   5 use crypto::digest::Digest;
   6 use crypto::sha2::Sha512Trunc256;
   7 use std::sync::Mutex;
   8
   9 use std::fs::{File, OpenOptions};
  10 use nix::fcntl::{flock, FlockArg};
  11 use std::os::unix::io::AsRawFd;
  12
  13 pub struct ChunkStore {
  14     base: PathBuf,
  15     chunk_dir: PathBuf,
  16     hasher: Sha512Trunc256,
  17     mutex: Mutex<bool>,
  18     lockfile: File,
  19 }
  20
  21 const HEX_CHARS: &'static [u8; 16] = b"0123456789abcdef";
  22
  23 pub fn digest_to_hex(digest: &[u8]) -> String {
  24
  25     let mut buf = Vec::<u8>::with_capacity(digest.len()*2);
  26
  27     for i in 0..digest.len() {
  28         buf.push(HEX_CHARS[(digest[i] >> 4) as usize]);
  29         buf.push(HEX_CHARS[(digest[i] & 0xf) as usize]);
  30     }
  31
  32     unsafe { String::from_utf8_unchecked(buf) }
  33 }
  34
  35 fn digest_to_prefix(digest: &[u8]) -> PathBuf {
  36
  37     let mut buf = Vec::<u8>::with_capacity(3+1+2+1);
  38
  39     buf.push(HEX_CHARS[(digest[0] as usize) >> 4]);
  40     buf.push(HEX_CHARS[(digest[0] as usize) &0xf]);
  41     buf.push(HEX_CHARS[(digest[1] as usize) >> 4]);
  42     buf.push('/' as u8);
  43
  44     buf.push(HEX_CHARS[(digest[1] as usize) & 0xf]);
  45     buf.push(HEX_CHARS[(digest[2] as usize) >> 4]);
  46     buf.push('/' as u8);
  47
  48     let path = unsafe { String::from_utf8_unchecked(buf)};
  49
  50     path.into()
  51 }
  52
  53 fn lock_file<P: AsRef<Path>>(filename: P, timeout: usize) -> Result<File, Error> {
  54
  55     let path = filename.as_ref();
  56     let lockfile = match OpenOptions::new()
  57         .create(true)
  58         .append(true)
  59         .open(path) {
  60             Ok(file) => file,
  61             Err(err) => bail!("Unable to open lock {:?} - {}",
  62                               path, err),
  63         };
  64
  65     let fd = lockfile.as_raw_fd();
  66
  67     let now = std::time::SystemTime::now();
  68     let mut print_msg = true;
  69     loop {
  70         match flock(fd, FlockArg::LockExclusiveNonblock) {
  71             Ok(_) => break,
  72             Err(_) => {
  73                 if print_msg {
  74                     print_msg = false;
  75                     eprintln!("trying to aquire lock...");
  76                 }
  77             }
  78         }
  79
  80         match now.elapsed() {
  81             Ok(elapsed) => {
  82                 if elapsed.as_secs() >= (timeout as u64) {
  83                     bail!("unable to aquire lock {:?} - got timeout", path);
  84                 }
  85             }
  86             Err(err) => {
  87                 bail!("unable to aquire lock {:?} - clock problems - {}", path, err);
  88             }
  89         }
  90         std::thread::sleep(std::time::Duration::from_millis(100));
  91     }
  92     Ok(lockfile)
  93 }
  94
  95 impl ChunkStore {
  96
  97     fn chunk_dir<P: AsRef<Path>>(path: P) -> PathBuf {
  98
  99         let mut chunk_dir: PathBuf = PathBuf::from(path.as_ref());
 100         chunk_dir.push(".chunks");
 101
 102         chunk_dir
 103     }
 104
 105     pub fn create<P: Into<PathBuf>>(path: P) -> Result<Self, Error> {
 106
 107         let base: PathBuf = path.into();
 108         let chunk_dir = Self::chunk_dir(&base);
 109
 110         if let Err(err) = std::fs::create_dir(&base) {
 111             bail!("unable to create chunk store {:?} - {}", base, err);
 112         }
 113
 114         if let Err(err) = std::fs::create_dir(&chunk_dir) {
 115             bail!("unable to create chunk store subdir {:?} - {}", chunk_dir, err);
 116         }
 117
 118         // create 4096 subdir
 119         for i in 0..4096 {
 120             let mut l1path = chunk_dir.clone();
 121             l1path.push(format!("{:03x}",i));
 122             if let Err(err) = std::fs::create_dir(&l1path) {
 123                 bail!("unable to create chunk subdir {:?} - {}", l1path, err);
 124             }
 125         }
 126
 127         Self::open(base)
 128     }
 129
 130     pub fn open<P: Into<PathBuf>>(path: P) -> Result<Self, Error> {
 131
 132         let base: PathBuf = path.into();
 133         let chunk_dir = Self::chunk_dir(&base);
 134
 135         if let Err(err) = std::fs::metadata(&chunk_dir) {
 136             bail!("unable to open chunk store {:?} - {}", chunk_dir, err);
 137         }
 138
 139         let mut lockfile_path = base.clone();
 140         lockfile_path.push(".lock");
 141
 142         // make sure only one process/thread/task can use it
 143         let lockfile = lock_file(lockfile_path, 10)?;
 144
 145         Ok(ChunkStore {
 146             base,
 147             chunk_dir,
 148             hasher: Sha512Trunc256::new(),
 149             lockfile,
 150             mutex: Mutex::new(false)
 151         })
 152     }
 153
 154     pub fn touch_chunk(&mut self, digest:&[u8]) ->  Result<(), Error> {
 155
 156         // fixme:  nix::sys::stat::utimensat
 157         let mut chunk_path = self.chunk_dir.clone();
 158         let prefix = digest_to_prefix(&digest);
 159         chunk_path.push(&prefix);
 160         let digest_str = digest_to_hex(&digest);
 161         chunk_path.push(&digest_str);
 162
 163         std::fs::metadata(&chunk_path)?;
 164         Ok(())
 165     }
 166
 167     fn sweep_old_files(&self, dir: &Path) {
 168
 169         let mut handle = match nix::dir::Dir::open(
 170             dir, nix::fcntl::OFlag::O_RDONLY, nix::sys::stat::Mode::empty()) {
 171             Ok(h) => h,
 172             Err(_) => return,
 173         };
 174
 175         let rawfd = handle.as_raw_fd();
 176
 177         let now = unsafe { libc::time(std::ptr::null_mut()) };
 178
 179         for entry in handle.iter() {
 180              match entry {
 181                 Ok(entry) => {
 182                     if let Some(file_type) = entry.file_type() {
 183                         if file_type == nix::dir::Type::File {
 184                             let filename = entry.file_name();
 185                             if let Ok(stat) = nix::sys::stat::fstatat(rawfd, filename, nix::fcntl::AtFlags::AT_SYMLINK_NOFOLLOW) {
 186                                 let age = now - stat.st_atime;
 187                                 println!("FOUND {}  {:?}", age/(3600*24), filename);
 188                                 if age/(3600*24) >= 2 {
 189                                     println!("UNLINK {}  {:?}", age/(3600*24), filename);
 190                                     unsafe { libc::unlinkat(rawfd, filename.as_ptr(), 0); }
 191                                 }
 192                             }
 193                         }
 194                     }
 195                 }
 196                 Err(_) => {
 197                     // fixme ??
 198                 }
 199              }
 200          }
 201
 202     }
 203
 204     pub fn sweep_used_chunks(&mut self) -> Result<(), Error> {
 205
 206         for i in 0..4096 {
 207             let mut l1path = self.chunk_dir.clone();
 208             l1path.push(format!("{:03x}", i));
 209             for j in 0..256 {
 210                 let mut l2path = l1path.clone();
 211                 l2path.push(format!("{:02x}", j));
 212                 self.sweep_old_files(&l2path);
 213             }
 214         }
 215
 216         Ok(())
 217     }
 218
 219     pub fn insert_chunk(&mut self, chunk: &[u8]) -> Result<(bool, [u8; 32]), Error> {
 220
 221         self.hasher.reset();
 222         self.hasher.input(chunk);
 223
 224         let mut digest = [0u8; 32];
 225         self.hasher.result(&mut digest);
 226         //println!("DIGEST {}", digest_to_hex(&digest));
 227
 228         let mut chunk_path = self.chunk_dir.clone();
 229         let prefix = digest_to_prefix(&digest);
 230         chunk_path.push(&prefix);
 231         let digest_str = digest_to_hex(&digest);
 232         chunk_path.push(&digest_str);
 233
 234         let lock = self.mutex.lock();
 235
 236         if let Ok(metadata) = std::fs::metadata(&chunk_path) {
 237             if metadata.is_file() {
 238                  return Ok((true, digest));
 239             } else {
 240                 bail!("Got unexpected file type for chunk {}", digest_str);
 241             }
 242         }
 243
 244         let mut chunk_dir = self.chunk_dir.clone();
 245         chunk_dir.push(&prefix);
 246
 247         if let Err(_) = std::fs::create_dir(&chunk_dir) { /* ignore */ }
 248
 249         let mut tmp_path = chunk_path.clone();
 250         tmp_path.set_extension("tmp");
 251         let mut f = std::fs::File::create(&tmp_path)?;
 252         f.write_all(chunk)?;
 253
 254         if let Err(err) = std::fs::rename(&tmp_path, &chunk_path) {
 255             if let Err(_) = std::fs::remove_file(&tmp_path)  { /* ignore */ }
 256             bail!("Atomic rename failed for chunk {} - {}", digest_str, err);
 257         }
 258
 259         println!("PATH {:?}", chunk_path);
 260
 261         drop(lock);
 262
 263         Ok((false, digest))
 264     }
 265
 266     pub fn relative_path(&self, path: &Path) -> PathBuf {
 267
 268         let mut full_path = self.base.clone();
 269         full_path.push(path);
 270         full_path
 271     }
 272
 273     pub fn base_path(&self) -> PathBuf {
 274         self.base.clone()
 275     }
 276
 277 }
 278
 279
 280 #[test]
 281 fn test_chunk_store1() {
 282
 283     if let Err(_e) = std::fs::remove_dir_all(".testdir") { /* ignore */ }
 284
 285     let chunk_store = ChunkStore::open(".testdir");
 286     assert!(chunk_store.is_err());
 287
 288     let mut chunk_store = ChunkStore::create(".testdir").unwrap();
 289     let (exists, _) = chunk_store.insert_chunk(&[0u8, 1u8]).unwrap();
 290     assert!(!exists);
 291
 292     let (exists, _) = chunk_store.insert_chunk(&[0u8, 1u8]).unwrap();
 293     assert!(exists);
 294
 295
 296     let chunk_store = ChunkStore::create(".testdir");
 297     assert!(chunk_store.is_err());
 298
 299
 300 }