]>
Commit | Line | Data |
---|---|---|
606ce64b DM |
1 | use failure::*; |
2 | ||
22968600 | 3 | use crate::tools; |
7bc1d727 | 4 | use super::IndexFile; |
7e336555 | 5 | use super::chunk_stat::*; |
606ce64b DM |
6 | use super::chunk_store::*; |
7 | ||
150f1bd8 | 8 | use std::sync::Arc; |
4818c8b6 | 9 | use std::io::{Read, Write}; |
10eea49d | 10 | use std::fs::File; |
606ce64b DM |
11 | use std::path::{Path, PathBuf}; |
12 | use std::os::unix::io::AsRawFd; | |
d13e3745 | 13 | use uuid::Uuid; |
4818c8b6 | 14 | use chrono::{Local, TimeZone}; |
f98ac774 | 15 | use super::ChunkInfo; |
606ce64b | 16 | |
8e39232a | 17 | /// Header format definition for fixed index files (`.fidx`) |
d13e3745 | 18 | #[repr(C)] |
91a905b6 | 19 | pub struct FixedIndexHeader { |
a7dd4830 | 20 | pub magic: [u8; 8], |
d13e3745 | 21 | pub uuid: [u8; 16], |
5e5b7f1c | 22 | pub ctime: u64, |
9335d74e DM |
23 | /// Sha256 over the index ``SHA256(digest1||digest2||...)`` |
24 | pub index_csum: [u8; 32], | |
a7dd4830 DM |
25 | pub size: u64, |
26 | pub chunk_size: u64, | |
27 | reserved: [u8; 4016], // overall size is one page (4096 bytes) | |
d13e3745 | 28 | } |
606ce64b DM |
29 | |
30 | // split image into fixed size chunks | |
31 | ||
91a905b6 | 32 | pub struct FixedIndexReader { |
150f1bd8 | 33 | store: Arc<ChunkStore>, |
10eea49d | 34 | _file: File, |
4818c8b6 | 35 | filename: PathBuf, |
29ae5c86 | 36 | pub chunk_size: usize, |
9f49fe1d | 37 | pub size: usize, |
e1225de4 | 38 | index_length: usize, |
4818c8b6 | 39 | index: *mut u8, |
9f49fe1d DM |
40 | pub uuid: [u8; 16], |
41 | pub ctime: u64, | |
9335d74e | 42 | pub index_csum: [u8; 32], |
4818c8b6 DM |
43 | } |
44 | ||
5be4065b WB |
45 | // `index` is mmap()ed which cannot be thread-local so should be sendable |
46 | unsafe impl Send for FixedIndexReader {} | |
47 | ||
91a905b6 | 48 | impl Drop for FixedIndexReader { |
4818c8b6 DM |
49 | |
50 | fn drop(&mut self) { | |
51 | if let Err(err) = self.unmap() { | |
52 | eprintln!("Unable to unmap file {:?} - {}", self.filename, err); | |
53 | } | |
54 | } | |
55 | } | |
56 | ||
91a905b6 | 57 | impl FixedIndexReader { |
4818c8b6 | 58 | |
150f1bd8 | 59 | pub fn open(store: Arc<ChunkStore>, path: &Path) -> Result<Self, Error> { |
4818c8b6 DM |
60 | |
61 | let full_path = store.relative_path(path); | |
62 | ||
877dd111 DM |
63 | let mut file = File::open(&full_path) |
64 | .map_err(|err| format_err!("Unable to open fixed index {:?} - {}", full_path, err))?; | |
4818c8b6 | 65 | |
c597a92c DM |
66 | if let Err(err) = nix::fcntl::flock(file.as_raw_fd(), nix::fcntl::FlockArg::LockSharedNonblock) { |
67 | bail!("unable to get shared lock on {:?} - {}", full_path, err); | |
68 | } | |
69 | ||
91a905b6 | 70 | let header_size = std::mem::size_of::<FixedIndexHeader>(); |
4818c8b6 DM |
71 | |
72 | // todo: use static assertion when available in rust | |
a360f6fa | 73 | if header_size != 4096 { bail!("got unexpected header size for {:?}", path); } |
4818c8b6 DM |
74 | |
75 | let mut buffer = vec![0u8; header_size]; | |
76 | file.read_exact(&mut buffer)?; | |
77 | ||
91a905b6 | 78 | let header = unsafe { &mut * (buffer.as_ptr() as *mut FixedIndexHeader) }; |
4818c8b6 | 79 | |
a7dd4830 | 80 | if header.magic != super::FIXED_SIZED_CHUNK_INDEX_1_0 { |
a360f6fa DM |
81 | bail!("got unknown magic number for {:?}", path); |
82 | } | |
83 | ||
48d0d356 DM |
84 | let size = u64::from_le(header.size) as usize; |
85 | let ctime = u64::from_le(header.ctime); | |
86 | let chunk_size = u64::from_le(header.chunk_size) as usize; | |
4818c8b6 | 87 | |
e1225de4 DM |
88 | let index_length = (size + chunk_size - 1)/chunk_size; |
89 | let index_size = index_length*32; | |
4818c8b6 | 90 | |
0b8e75ed DM |
91 | let rawfd = file.as_raw_fd(); |
92 | ||
93 | let stat = match nix::sys::stat::fstat(rawfd) { | |
94 | Ok(stat) => stat, | |
95 | Err(err) => bail!("fstat {:?} failed - {}", path, err), | |
96 | }; | |
97 | ||
ddbdf80d | 98 | let expected_index_size = (stat.st_size as usize) - header_size; |
0b8e75ed DM |
99 | if index_size != expected_index_size { |
100 | bail!("got unexpected file size for {:?} ({} != {})", | |
101 | path, index_size, expected_index_size); | |
102 | } | |
4818c8b6 DM |
103 | |
104 | let data = unsafe { nix::sys::mman::mmap( | |
105 | std::ptr::null_mut(), | |
106 | index_size, | |
107 | nix::sys::mman::ProtFlags::PROT_READ, | |
108 | nix::sys::mman::MapFlags::MAP_PRIVATE, | |
109 | file.as_raw_fd(), | |
110 | header_size as i64) }? as *mut u8; | |
111 | ||
112 | Ok(Self { | |
113 | store, | |
114 | filename: full_path, | |
10eea49d | 115 | _file: file, |
4818c8b6 DM |
116 | chunk_size, |
117 | size, | |
e1225de4 | 118 | index_length, |
4818c8b6 DM |
119 | index: data, |
120 | ctime, | |
121 | uuid: header.uuid, | |
9335d74e | 122 | index_csum: header.index_csum, |
4818c8b6 DM |
123 | }) |
124 | } | |
125 | ||
126 | fn unmap(&mut self) -> Result<(), Error> { | |
127 | ||
128 | if self.index == std::ptr::null_mut() { return Ok(()); } | |
129 | ||
e1225de4 | 130 | let index_size = self.index_length*32; |
4818c8b6 DM |
131 | |
132 | if let Err(err) = unsafe { nix::sys::mman::munmap(self.index as *mut std::ffi::c_void, index_size) } { | |
133 | bail!("unmap file {:?} failed - {}", self.filename, err); | |
134 | } | |
135 | ||
136 | self.index = std::ptr::null_mut(); | |
137 | ||
138 | Ok(()) | |
139 | } | |
140 | ||
141 | pub fn print_info(&self) { | |
142 | println!("Filename: {:?}", self.filename); | |
143 | println!("Size: {}", self.size); | |
144 | println!("ChunkSize: {}", self.chunk_size); | |
145 | println!("CTime: {}", Local.timestamp(self.ctime as i64, 0).format("%c")); | |
146 | println!("UUID: {:?}", self.uuid); | |
147 | } | |
148 | } | |
149 | ||
7bc1d727 WB |
150 | impl IndexFile for FixedIndexReader { |
151 | fn index_count(&self) -> usize { | |
e1225de4 | 152 | self.index_length |
7bc1d727 WB |
153 | } |
154 | ||
155 | fn index_digest(&self, pos: usize) -> Option<&[u8; 32]> { | |
e1225de4 | 156 | if pos >= self.index_length { |
7bc1d727 WB |
157 | None |
158 | } else { | |
159 | Some(unsafe { std::mem::transmute(self.index.add(pos*32)) }) | |
160 | } | |
161 | } | |
a660978c DM |
162 | |
163 | fn index_bytes(&self) -> u64 { | |
164 | (self.index_length * self.chunk_size) as u64 | |
165 | } | |
7bc1d727 WB |
166 | } |
167 | ||
91a905b6 | 168 | pub struct FixedIndexWriter { |
150f1bd8 | 169 | store: Arc<ChunkStore>, |
9335d74e | 170 | file: File, |
43b13033 | 171 | _lock: tools::ProcessLockSharedGuard, |
4fbb72a8 DM |
172 | filename: PathBuf, |
173 | tmp_filename: PathBuf, | |
606ce64b DM |
174 | chunk_size: usize, |
175 | size: usize, | |
e1225de4 | 176 | index_length: usize, |
606ce64b | 177 | index: *mut u8, |
9f49fe1d DM |
178 | pub uuid: [u8; 16], |
179 | pub ctime: u64, | |
606ce64b DM |
180 | } |
181 | ||
c3bb97e5 WB |
182 | // `index` is mmap()ed which cannot be thread-local so should be sendable |
183 | unsafe impl Send for FixedIndexWriter {} | |
184 | ||
91a905b6 | 185 | impl Drop for FixedIndexWriter { |
4fbb72a8 DM |
186 | |
187 | fn drop(&mut self) { | |
188 | let _ = std::fs::remove_file(&self.tmp_filename); // ignore errors | |
189 | if let Err(err) = self.unmap() { | |
0cd9d420 | 190 | eprintln!("Unable to unmap file {:?} - {}", self.tmp_filename, err); |
4fbb72a8 DM |
191 | } |
192 | } | |
193 | } | |
194 | ||
91a905b6 | 195 | impl FixedIndexWriter { |
606ce64b | 196 | |
150f1bd8 | 197 | pub fn create(store: Arc<ChunkStore>, path: &Path, size: usize, chunk_size: usize) -> Result<Self, Error> { |
606ce64b | 198 | |
43b13033 DM |
199 | let shared_lock = store.try_shared_lock()?; |
200 | ||
606ce64b | 201 | let full_path = store.relative_path(path); |
4fbb72a8 | 202 | let mut tmp_path = full_path.clone(); |
91a905b6 | 203 | tmp_path.set_extension("tmp_fidx"); |
606ce64b DM |
204 | |
205 | let mut file = std::fs::OpenOptions::new() | |
d13e3745 | 206 | .create(true).truncate(true) |
606ce64b DM |
207 | .read(true) |
208 | .write(true) | |
4fbb72a8 | 209 | .open(&tmp_path)?; |
606ce64b | 210 | |
91a905b6 | 211 | let header_size = std::mem::size_of::<FixedIndexHeader>(); |
d13e3745 DM |
212 | |
213 | // todo: use static assertion when available in rust | |
214 | if header_size != 4096 { panic!("got unexpected header size"); } | |
215 | ||
216 | let ctime = std::time::SystemTime::now().duration_since( | |
5e5b7f1c | 217 | std::time::SystemTime::UNIX_EPOCH)?.as_secs(); |
d13e3745 DM |
218 | |
219 | let uuid = Uuid::new_v4(); | |
220 | ||
0cd9d420 | 221 | let buffer = vec![0u8; header_size]; |
91a905b6 | 222 | let header = unsafe { &mut * (buffer.as_ptr() as *mut FixedIndexHeader) }; |
d13e3745 | 223 | |
a7dd4830 | 224 | header.magic = super::FIXED_SIZED_CHUNK_INDEX_1_0; |
48d0d356 DM |
225 | header.ctime = u64::to_le(ctime); |
226 | header.size = u64::to_le(size as u64); | |
227 | header.chunk_size = u64::to_le(chunk_size as u64); | |
d13e3745 DM |
228 | header.uuid = *uuid.as_bytes(); |
229 | ||
9335d74e DM |
230 | header.index_csum = [0u8; 32]; |
231 | ||
5e5b7f1c | 232 | file.write_all(&buffer)?; |
d13e3745 | 233 | |
e1225de4 DM |
234 | let index_length = (size + chunk_size - 1)/chunk_size; |
235 | let index_size = index_length*32; | |
d13e3745 DM |
236 | nix::unistd::ftruncate(file.as_raw_fd(), (header_size + index_size) as i64)?; |
237 | ||
606ce64b DM |
238 | let data = unsafe { nix::sys::mman::mmap( |
239 | std::ptr::null_mut(), | |
240 | index_size, | |
241 | nix::sys::mman::ProtFlags::PROT_READ | nix::sys::mman::ProtFlags::PROT_WRITE, | |
242 | nix::sys::mman::MapFlags::MAP_SHARED, | |
243 | file.as_raw_fd(), | |
d13e3745 DM |
244 | header_size as i64) }? as *mut u8; |
245 | ||
606ce64b DM |
246 | Ok(Self { |
247 | store, | |
9335d74e | 248 | file, |
43b13033 | 249 | _lock: shared_lock, |
4fbb72a8 DM |
250 | filename: full_path, |
251 | tmp_filename: tmp_path, | |
606ce64b DM |
252 | chunk_size, |
253 | size, | |
e1225de4 | 254 | index_length, |
606ce64b | 255 | index: data, |
d13e3745 DM |
256 | ctime, |
257 | uuid: *uuid.as_bytes(), | |
606ce64b DM |
258 | }) |
259 | } | |
260 | ||
006f3ff4 DM |
261 | pub fn index_length(&self) -> usize { |
262 | self.index_length | |
263 | } | |
264 | ||
4fbb72a8 DM |
265 | fn unmap(&mut self) -> Result<(), Error> { |
266 | ||
267 | if self.index == std::ptr::null_mut() { return Ok(()); } | |
268 | ||
e1225de4 | 269 | let index_size = self.index_length*32; |
4fbb72a8 DM |
270 | |
271 | if let Err(err) = unsafe { nix::sys::mman::munmap(self.index as *mut std::ffi::c_void, index_size) } { | |
0cd9d420 | 272 | bail!("unmap file {:?} failed - {}", self.tmp_filename, err); |
4fbb72a8 DM |
273 | } |
274 | ||
275 | self.index = std::ptr::null_mut(); | |
276 | ||
277 | Ok(()) | |
278 | } | |
279 | ||
9335d74e | 280 | pub fn close(&mut self) -> Result<[u8; 32], Error> { |
4fbb72a8 DM |
281 | |
282 | if self.index == std::ptr::null_mut() { bail!("cannot close already closed index file."); } | |
283 | ||
9335d74e DM |
284 | let index_size = self.index_length*32; |
285 | let data = unsafe { std::slice::from_raw_parts(self.index, index_size) }; | |
286 | let index_csum = openssl::sha::sha256(data); | |
287 | ||
4fbb72a8 DM |
288 | self.unmap()?; |
289 | ||
9335d74e DM |
290 | use std::io::Seek; |
291 | ||
292 | let csum_offset = proxmox::tools::offsetof!(FixedIndexHeader, index_csum); | |
293 | self.file.seek(std::io::SeekFrom::Start(csum_offset as u64))?; | |
294 | self.file.write_all(&index_csum)?; | |
295 | self.file.flush()?; | |
296 | ||
4fbb72a8 DM |
297 | if let Err(err) = std::fs::rename(&self.tmp_filename, &self.filename) { |
298 | bail!("Atomic rename file {:?} failed - {}", self.filename, err); | |
299 | } | |
300 | ||
9335d74e | 301 | Ok(index_csum) |
4fbb72a8 DM |
302 | } |
303 | ||
01af11f3 | 304 | // Note: We want to add data out of order, so do not assume any order here. |
f98ac774 | 305 | pub fn add_chunk(&mut self, chunk_info: &ChunkInfo, stat: &mut ChunkStat) -> Result<(), Error> { |
606ce64b | 306 | |
f98ac774 DM |
307 | let chunk_len = chunk_info.chunk_len as usize; |
308 | let end = chunk_info.offset as usize; | |
309 | ||
310 | if end < chunk_len { | |
311 | bail!("got chunk with small offset ({} < {}", end, chunk_len); | |
312 | } | |
313 | ||
314 | let pos = end - chunk_len; | |
606ce64b DM |
315 | |
316 | if end > self.size { | |
317 | bail!("write chunk data exceeds size ({} >= {})", end, self.size); | |
318 | } | |
319 | ||
320 | // last chunk can be smaller | |
f98ac774 DM |
321 | if ((end != self.size) && (chunk_len != self.chunk_size)) || |
322 | (chunk_len > self.chunk_size) || (chunk_len == 0) { | |
323 | bail!("got chunk with wrong length ({} != {}", chunk_len, self.chunk_size); | |
606ce64b DM |
324 | } |
325 | ||
606ce64b DM |
326 | if pos & (self.chunk_size-1) != 0 { bail!("add unaligned chunk (pos = {})", pos); } |
327 | ||
f98ac774 DM |
328 | if (end as u64) != chunk_info.offset { |
329 | bail!("got chunk with wrong offset ({} != {}", end, chunk_info.offset); | |
330 | } | |
331 | ||
332 | let (is_duplicate, compressed_size) = self.store.insert_chunk(&chunk_info.chunk)?; | |
798f7fa0 | 333 | |
cb0708dd DM |
334 | stat.chunk_count += 1; |
335 | stat.compressed_size += compressed_size; | |
606ce64b | 336 | |
f98ac774 DM |
337 | let digest = chunk_info.chunk.digest(); |
338 | ||
339 | println!("ADD CHUNK {} {} {}% {} {}", pos, chunk_len, | |
bffd40d6 | 340 | (compressed_size*100)/(chunk_len as u64), is_duplicate, proxmox::tools::digest_to_hex(digest)); |
798f7fa0 DM |
341 | |
342 | if is_duplicate { | |
cb0708dd | 343 | stat.duplicate_chunks += 1; |
798f7fa0 | 344 | } else { |
cb0708dd | 345 | stat.disk_size += compressed_size; |
798f7fa0 | 346 | } |
606ce64b | 347 | |
f98ac774 | 348 | self.add_digest(pos / self.chunk_size, digest) |
e3062f87 WB |
349 | } |
350 | ||
351 | pub fn add_digest(&mut self, index: usize, digest: &[u8; 32]) -> Result<(), Error> { | |
01af11f3 | 352 | |
fc14b849 DM |
353 | if index >= self.index_length { |
354 | bail!("add digest failed - index out of range ({} >= {})", index, self.index_length); | |
355 | } | |
356 | ||
01af11f3 DM |
357 | if self.index == std::ptr::null_mut() { bail!("cannot write to closed index file."); } |
358 | ||
e3062f87 | 359 | let index_pos = index*32; |
606ce64b DM |
360 | unsafe { |
361 | let dst = self.index.add(index_pos); | |
362 | dst.copy_from_nonoverlapping(digest.as_ptr(), 32); | |
363 | } | |
364 | ||
365 | Ok(()) | |
366 | } | |
367 | } |