2018-12-15 13:51:05 +00:00
|
|
|
use failure::*;
|
2019-07-05 07:19:56 +00:00
|
|
|
use std::io::{Seek, SeekFrom};
|
2018-12-15 13:51:05 +00:00
|
|
|
|
2019-01-25 09:58:28 +00:00
|
|
|
use crate::tools;
|
2019-02-27 13:32:34 +00:00
|
|
|
use super::IndexFile;
|
2019-02-25 11:52:10 +00:00
|
|
|
use super::chunk_stat::*;
|
2018-12-15 13:51:05 +00:00
|
|
|
use super::chunk_store::*;
|
|
|
|
|
2019-01-15 11:36:16 +00:00
|
|
|
use std::sync::Arc;
|
2018-12-16 13:44:44 +00:00
|
|
|
use std::io::{Read, Write};
|
2019-03-27 09:17:03 +00:00
|
|
|
use std::fs::File;
|
2018-12-15 13:51:05 +00:00
|
|
|
use std::path::{Path, PathBuf};
|
|
|
|
use std::os::unix::io::AsRawFd;
|
2018-12-15 16:05:49 +00:00
|
|
|
use uuid::Uuid;
|
2018-12-16 13:44:44 +00:00
|
|
|
use chrono::{Local, TimeZone};
|
2019-07-05 07:19:56 +00:00
|
|
|
|
2019-06-13 09:47:23 +00:00
|
|
|
use super::ChunkInfo;
|
2019-07-05 07:19:56 +00:00
|
|
|
use super::read_chunk::*;
|
2018-12-15 13:51:05 +00:00
|
|
|
|
2019-03-11 09:51:48 +00:00
|
|
|
/// Header format definition for fixed index files (`.fidx`)
|
2018-12-15 16:05:49 +00:00
|
|
|
#[repr(C)]
|
2019-02-12 10:50:45 +00:00
|
|
|
pub struct FixedIndexHeader {
|
2019-06-14 12:58:37 +00:00
|
|
|
pub magic: [u8; 8],
|
2018-12-15 16:05:49 +00:00
|
|
|
pub uuid: [u8; 16],
|
2018-12-16 10:48:03 +00:00
|
|
|
pub ctime: u64,
|
2019-06-14 08:34:29 +00:00
|
|
|
/// Sha256 over the index ``SHA256(digest1||digest2||...)``
|
|
|
|
pub index_csum: [u8; 32],
|
2019-06-14 12:58:37 +00:00
|
|
|
pub size: u64,
|
|
|
|
pub chunk_size: u64,
|
|
|
|
reserved: [u8; 4016], // overall size is one page (4096 bytes)
|
2018-12-15 16:05:49 +00:00
|
|
|
}
|
2018-12-15 13:51:05 +00:00
|
|
|
|
|
|
|
// split image into fixed size chunks
|
|
|
|
|
2019-02-12 10:50:45 +00:00
|
|
|
pub struct FixedIndexReader {
|
2019-03-27 09:17:03 +00:00
|
|
|
_file: File,
|
2019-05-28 04:18:24 +00:00
|
|
|
pub chunk_size: usize,
|
2019-07-04 13:13:22 +00:00
|
|
|
pub size: u64,
|
2019-05-28 06:27:57 +00:00
|
|
|
index_length: usize,
|
2018-12-16 13:44:44 +00:00
|
|
|
index: *mut u8,
|
2019-01-30 17:25:37 +00:00
|
|
|
pub uuid: [u8; 16],
|
|
|
|
pub ctime: u64,
|
2019-06-14 08:34:29 +00:00
|
|
|
pub index_csum: [u8; 32],
|
2018-12-16 13:44:44 +00:00
|
|
|
}
|
|
|
|
|
2019-02-28 08:36:19 +00:00
|
|
|
// `index` is mmap()ed which cannot be thread-local so should be sendable
|
|
|
|
unsafe impl Send for FixedIndexReader {}
|
|
|
|
|
2019-02-12 10:50:45 +00:00
|
|
|
impl Drop for FixedIndexReader {
|
2018-12-16 13:44:44 +00:00
|
|
|
|
|
|
|
fn drop(&mut self) {
|
|
|
|
if let Err(err) = self.unmap() {
|
2019-07-04 06:09:48 +00:00
|
|
|
eprintln!("Unable to unmap file - {}", err);
|
2018-12-16 13:44:44 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2019-02-12 10:50:45 +00:00
|
|
|
impl FixedIndexReader {
|
2018-12-16 13:44:44 +00:00
|
|
|
|
2019-07-04 06:09:48 +00:00
|
|
|
pub fn open(path: &Path) -> Result<Self, Error> {
|
2018-12-16 13:44:44 +00:00
|
|
|
|
2019-07-04 06:09:48 +00:00
|
|
|
File::open(path)
|
|
|
|
.map_err(Error::from)
|
|
|
|
.and_then(|file| Self::new(file))
|
|
|
|
.map_err(|err| format_err!("Unable to open fixed index {:?} - {}", path, err))
|
|
|
|
}
|
2018-12-16 13:44:44 +00:00
|
|
|
|
2019-07-04 06:09:48 +00:00
|
|
|
pub fn new(mut file: std::fs::File) -> Result<Self, Error> {
|
2018-12-16 13:44:44 +00:00
|
|
|
|
2019-03-27 09:11:14 +00:00
|
|
|
if let Err(err) = nix::fcntl::flock(file.as_raw_fd(), nix::fcntl::FlockArg::LockSharedNonblock) {
|
2019-07-04 06:09:48 +00:00
|
|
|
bail!("unable to get shared lock - {}", err);
|
2019-03-27 09:11:14 +00:00
|
|
|
}
|
|
|
|
|
2019-07-05 07:19:56 +00:00
|
|
|
file.seek(SeekFrom::Start(0))?;
|
|
|
|
|
2019-02-12 10:50:45 +00:00
|
|
|
let header_size = std::mem::size_of::<FixedIndexHeader>();
|
2018-12-16 13:44:44 +00:00
|
|
|
|
|
|
|
// todo: use static assertion when available in rust
|
2019-07-04 06:09:48 +00:00
|
|
|
if header_size != 4096 { bail!("got unexpected header size"); }
|
2018-12-16 13:44:44 +00:00
|
|
|
|
|
|
|
let mut buffer = vec![0u8; header_size];
|
|
|
|
file.read_exact(&mut buffer)?;
|
|
|
|
|
2019-02-12 10:50:45 +00:00
|
|
|
let header = unsafe { &mut * (buffer.as_ptr() as *mut FixedIndexHeader) };
|
2018-12-16 13:44:44 +00:00
|
|
|
|
2019-06-14 12:58:37 +00:00
|
|
|
if header.magic != super::FIXED_SIZED_CHUNK_INDEX_1_0 {
|
2019-07-04 06:09:48 +00:00
|
|
|
bail!("got unknown magic number");
|
2019-01-02 12:13:13 +00:00
|
|
|
}
|
|
|
|
|
2019-07-04 13:13:22 +00:00
|
|
|
let size = u64::from_le(header.size);
|
2019-01-04 07:45:04 +00:00
|
|
|
let ctime = u64::from_le(header.ctime);
|
2019-07-04 13:13:22 +00:00
|
|
|
let chunk_size = u64::from_le(header.chunk_size);
|
2018-12-16 13:44:44 +00:00
|
|
|
|
2019-07-04 13:13:22 +00:00
|
|
|
let index_length = ((size + chunk_size - 1)/chunk_size) as usize;
|
2019-05-28 06:27:57 +00:00
|
|
|
let index_size = index_length*32;
|
2018-12-16 13:44:44 +00:00
|
|
|
|
2019-01-02 17:14:02 +00:00
|
|
|
let rawfd = file.as_raw_fd();
|
|
|
|
|
|
|
|
let stat = match nix::sys::stat::fstat(rawfd) {
|
|
|
|
Ok(stat) => stat,
|
2019-07-04 06:09:48 +00:00
|
|
|
Err(err) => bail!("fstat failed - {}", err),
|
2019-01-02 17:14:02 +00:00
|
|
|
};
|
|
|
|
|
2019-01-11 07:41:33 +00:00
|
|
|
let expected_index_size = (stat.st_size as usize) - header_size;
|
2019-01-02 17:14:02 +00:00
|
|
|
if index_size != expected_index_size {
|
2019-07-04 06:09:48 +00:00
|
|
|
bail!("got unexpected file size ({} != {})", index_size, expected_index_size);
|
2019-01-02 17:14:02 +00:00
|
|
|
}
|
2018-12-16 13:44:44 +00:00
|
|
|
|
|
|
|
let data = unsafe { nix::sys::mman::mmap(
|
|
|
|
std::ptr::null_mut(),
|
|
|
|
index_size,
|
|
|
|
nix::sys::mman::ProtFlags::PROT_READ,
|
|
|
|
nix::sys::mman::MapFlags::MAP_PRIVATE,
|
|
|
|
file.as_raw_fd(),
|
|
|
|
header_size as i64) }? as *mut u8;
|
|
|
|
|
|
|
|
Ok(Self {
|
2019-03-27 09:17:03 +00:00
|
|
|
_file: file,
|
2019-07-04 13:13:22 +00:00
|
|
|
chunk_size: chunk_size as usize,
|
2018-12-16 13:44:44 +00:00
|
|
|
size,
|
2019-05-28 06:27:57 +00:00
|
|
|
index_length,
|
2018-12-16 13:44:44 +00:00
|
|
|
index: data,
|
|
|
|
ctime,
|
|
|
|
uuid: header.uuid,
|
2019-06-14 08:34:29 +00:00
|
|
|
index_csum: header.index_csum,
|
2018-12-16 13:44:44 +00:00
|
|
|
})
|
|
|
|
}
|
|
|
|
|
|
|
|
fn unmap(&mut self) -> Result<(), Error> {
|
|
|
|
|
|
|
|
if self.index == std::ptr::null_mut() { return Ok(()); }
|
|
|
|
|
2019-05-28 06:27:57 +00:00
|
|
|
let index_size = self.index_length*32;
|
2018-12-16 13:44:44 +00:00
|
|
|
|
|
|
|
if let Err(err) = unsafe { nix::sys::mman::munmap(self.index as *mut std::ffi::c_void, index_size) } {
|
2019-07-04 06:09:48 +00:00
|
|
|
bail!("unmap file failed - {}", err);
|
2018-12-16 13:44:44 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
self.index = std::ptr::null_mut();
|
|
|
|
|
|
|
|
Ok(())
|
|
|
|
}
|
|
|
|
|
2019-07-05 07:19:56 +00:00
|
|
|
pub fn chunk_info(&self, pos: usize) -> Result<(u64, u64, [u8; 32]), Error> {
|
|
|
|
|
|
|
|
if pos >= self.index_length {
|
|
|
|
bail!("chunk index out of range");
|
|
|
|
}
|
|
|
|
let start = (pos * self.chunk_size) as u64;
|
|
|
|
let mut end = start + self.chunk_size as u64;
|
|
|
|
|
|
|
|
if end > self.size {
|
|
|
|
end = self.size;
|
|
|
|
}
|
|
|
|
|
|
|
|
let mut digest: [u8; 32] = unsafe { std::mem::uninitialized() };
|
|
|
|
unsafe { std::ptr::copy_nonoverlapping(self.index.add(pos*32), digest.as_mut_ptr(), 32); }
|
|
|
|
|
|
|
|
Ok((start, end, digest))
|
|
|
|
}
|
|
|
|
|
|
|
|
#[inline]
|
|
|
|
fn chunk_end(&self, pos: usize) -> u64 {
|
|
|
|
if pos >= self.index_length {
|
|
|
|
panic!("chunk index out of range");
|
|
|
|
}
|
|
|
|
|
|
|
|
let end = ((pos+1) * self.chunk_size) as u64;
|
|
|
|
if end > self.size {
|
|
|
|
self.size
|
|
|
|
} else {
|
|
|
|
end
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2018-12-16 13:44:44 +00:00
|
|
|
pub fn print_info(&self) {
|
|
|
|
println!("Size: {}", self.size);
|
|
|
|
println!("ChunkSize: {}", self.chunk_size);
|
|
|
|
println!("CTime: {}", Local.timestamp(self.ctime as i64, 0).format("%c"));
|
|
|
|
println!("UUID: {:?}", self.uuid);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2019-02-27 13:32:34 +00:00
|
|
|
impl IndexFile for FixedIndexReader {
|
|
|
|
fn index_count(&self) -> usize {
|
2019-05-28 06:27:57 +00:00
|
|
|
self.index_length
|
2019-02-27 13:32:34 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
fn index_digest(&self, pos: usize) -> Option<&[u8; 32]> {
|
2019-05-28 06:27:57 +00:00
|
|
|
if pos >= self.index_length {
|
2019-02-27 13:32:34 +00:00
|
|
|
None
|
|
|
|
} else {
|
|
|
|
Some(unsafe { std::mem::transmute(self.index.add(pos*32)) })
|
|
|
|
}
|
|
|
|
}
|
2019-07-04 05:57:43 +00:00
|
|
|
|
|
|
|
fn index_bytes(&self) -> u64 {
|
2019-07-04 13:13:22 +00:00
|
|
|
self.size
|
2019-07-04 05:57:43 +00:00
|
|
|
}
|
2019-02-27 13:32:34 +00:00
|
|
|
}
|
|
|
|
|
2019-02-12 10:50:45 +00:00
|
|
|
pub struct FixedIndexWriter {
|
2019-01-15 11:36:16 +00:00
|
|
|
store: Arc<ChunkStore>,
|
2019-06-14 08:34:29 +00:00
|
|
|
file: File,
|
2019-03-22 08:42:15 +00:00
|
|
|
_lock: tools::ProcessLockSharedGuard,
|
2018-12-16 12:39:21 +00:00
|
|
|
filename: PathBuf,
|
|
|
|
tmp_filename: PathBuf,
|
2018-12-15 13:51:05 +00:00
|
|
|
chunk_size: usize,
|
|
|
|
size: usize,
|
2019-05-28 06:27:57 +00:00
|
|
|
index_length: usize,
|
2018-12-15 13:51:05 +00:00
|
|
|
index: *mut u8,
|
2019-01-30 17:25:37 +00:00
|
|
|
pub uuid: [u8; 16],
|
|
|
|
pub ctime: u64,
|
2018-12-15 13:51:05 +00:00
|
|
|
}
|
|
|
|
|
2019-03-06 09:20:04 +00:00
|
|
|
// `index` is mmap()ed which cannot be thread-local so should be sendable
|
|
|
|
unsafe impl Send for FixedIndexWriter {}
|
|
|
|
|
2019-02-12 10:50:45 +00:00
|
|
|
impl Drop for FixedIndexWriter {
|
2018-12-16 12:39:21 +00:00
|
|
|
|
|
|
|
fn drop(&mut self) {
|
|
|
|
let _ = std::fs::remove_file(&self.tmp_filename); // ignore errors
|
|
|
|
if let Err(err) = self.unmap() {
|
2018-12-16 12:43:19 +00:00
|
|
|
eprintln!("Unable to unmap file {:?} - {}", self.tmp_filename, err);
|
2018-12-16 12:39:21 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2019-02-12 10:50:45 +00:00
|
|
|
impl FixedIndexWriter {
|
2018-12-15 13:51:05 +00:00
|
|
|
|
2019-01-15 11:36:16 +00:00
|
|
|
pub fn create(store: Arc<ChunkStore>, path: &Path, size: usize, chunk_size: usize) -> Result<Self, Error> {
|
2018-12-15 13:51:05 +00:00
|
|
|
|
2019-03-22 08:42:15 +00:00
|
|
|
let shared_lock = store.try_shared_lock()?;
|
|
|
|
|
2018-12-15 13:51:05 +00:00
|
|
|
let full_path = store.relative_path(path);
|
2018-12-16 12:39:21 +00:00
|
|
|
let mut tmp_path = full_path.clone();
|
2019-02-12 10:50:45 +00:00
|
|
|
tmp_path.set_extension("tmp_fidx");
|
2018-12-15 13:51:05 +00:00
|
|
|
|
|
|
|
let mut file = std::fs::OpenOptions::new()
|
2018-12-15 16:05:49 +00:00
|
|
|
.create(true).truncate(true)
|
2018-12-15 13:51:05 +00:00
|
|
|
.read(true)
|
|
|
|
.write(true)
|
2018-12-16 12:39:21 +00:00
|
|
|
.open(&tmp_path)?;
|
2018-12-15 13:51:05 +00:00
|
|
|
|
2019-02-12 10:50:45 +00:00
|
|
|
let header_size = std::mem::size_of::<FixedIndexHeader>();
|
2018-12-15 16:05:49 +00:00
|
|
|
|
|
|
|
// todo: use static assertion when available in rust
|
|
|
|
if header_size != 4096 { panic!("got unexpected header size"); }
|
|
|
|
|
|
|
|
let ctime = std::time::SystemTime::now().duration_since(
|
2018-12-16 10:48:03 +00:00
|
|
|
std::time::SystemTime::UNIX_EPOCH)?.as_secs();
|
2018-12-15 16:05:49 +00:00
|
|
|
|
|
|
|
let uuid = Uuid::new_v4();
|
|
|
|
|
2018-12-16 12:43:19 +00:00
|
|
|
let buffer = vec![0u8; header_size];
|
2019-02-12 10:50:45 +00:00
|
|
|
let header = unsafe { &mut * (buffer.as_ptr() as *mut FixedIndexHeader) };
|
2018-12-15 16:05:49 +00:00
|
|
|
|
2019-06-14 12:58:37 +00:00
|
|
|
header.magic = super::FIXED_SIZED_CHUNK_INDEX_1_0;
|
2019-01-04 07:45:04 +00:00
|
|
|
header.ctime = u64::to_le(ctime);
|
|
|
|
header.size = u64::to_le(size as u64);
|
|
|
|
header.chunk_size = u64::to_le(chunk_size as u64);
|
2018-12-15 16:05:49 +00:00
|
|
|
header.uuid = *uuid.as_bytes();
|
|
|
|
|
2019-06-14 08:34:29 +00:00
|
|
|
header.index_csum = [0u8; 32];
|
|
|
|
|
2018-12-16 10:48:03 +00:00
|
|
|
file.write_all(&buffer)?;
|
2018-12-15 16:05:49 +00:00
|
|
|
|
2019-05-28 06:27:57 +00:00
|
|
|
let index_length = (size + chunk_size - 1)/chunk_size;
|
|
|
|
let index_size = index_length*32;
|
2018-12-15 16:05:49 +00:00
|
|
|
nix::unistd::ftruncate(file.as_raw_fd(), (header_size + index_size) as i64)?;
|
|
|
|
|
2018-12-15 13:51:05 +00:00
|
|
|
let data = unsafe { nix::sys::mman::mmap(
|
|
|
|
std::ptr::null_mut(),
|
|
|
|
index_size,
|
|
|
|
nix::sys::mman::ProtFlags::PROT_READ | nix::sys::mman::ProtFlags::PROT_WRITE,
|
|
|
|
nix::sys::mman::MapFlags::MAP_SHARED,
|
|
|
|
file.as_raw_fd(),
|
2018-12-15 16:05:49 +00:00
|
|
|
header_size as i64) }? as *mut u8;
|
|
|
|
|
2018-12-15 13:51:05 +00:00
|
|
|
Ok(Self {
|
|
|
|
store,
|
2019-06-14 08:34:29 +00:00
|
|
|
file,
|
2019-03-22 08:42:15 +00:00
|
|
|
_lock: shared_lock,
|
2018-12-16 12:39:21 +00:00
|
|
|
filename: full_path,
|
|
|
|
tmp_filename: tmp_path,
|
2018-12-15 13:51:05 +00:00
|
|
|
chunk_size,
|
|
|
|
size,
|
2019-05-28 06:27:57 +00:00
|
|
|
index_length,
|
2018-12-15 13:51:05 +00:00
|
|
|
index: data,
|
2018-12-15 16:05:49 +00:00
|
|
|
ctime,
|
|
|
|
uuid: *uuid.as_bytes(),
|
2018-12-15 13:51:05 +00:00
|
|
|
})
|
|
|
|
}
|
|
|
|
|
2019-05-28 07:12:38 +00:00
|
|
|
pub fn index_length(&self) -> usize {
|
|
|
|
self.index_length
|
|
|
|
}
|
|
|
|
|
2018-12-16 12:39:21 +00:00
|
|
|
fn unmap(&mut self) -> Result<(), Error> {
|
|
|
|
|
|
|
|
if self.index == std::ptr::null_mut() { return Ok(()); }
|
|
|
|
|
2019-05-28 06:27:57 +00:00
|
|
|
let index_size = self.index_length*32;
|
2018-12-16 12:39:21 +00:00
|
|
|
|
|
|
|
if let Err(err) = unsafe { nix::sys::mman::munmap(self.index as *mut std::ffi::c_void, index_size) } {
|
2018-12-16 12:43:19 +00:00
|
|
|
bail!("unmap file {:?} failed - {}", self.tmp_filename, err);
|
2018-12-16 12:39:21 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
self.index = std::ptr::null_mut();
|
|
|
|
|
|
|
|
Ok(())
|
|
|
|
}
|
|
|
|
|
2019-06-14 08:34:29 +00:00
|
|
|
pub fn close(&mut self) -> Result<[u8; 32], Error> {
|
2018-12-16 12:39:21 +00:00
|
|
|
|
|
|
|
if self.index == std::ptr::null_mut() { bail!("cannot close already closed index file."); }
|
|
|
|
|
2019-06-14 08:34:29 +00:00
|
|
|
let index_size = self.index_length*32;
|
|
|
|
let data = unsafe { std::slice::from_raw_parts(self.index, index_size) };
|
|
|
|
let index_csum = openssl::sha::sha256(data);
|
|
|
|
|
2018-12-16 12:39:21 +00:00
|
|
|
self.unmap()?;
|
|
|
|
|
2019-06-14 08:34:29 +00:00
|
|
|
let csum_offset = proxmox::tools::offsetof!(FixedIndexHeader, index_csum);
|
2019-07-05 07:19:56 +00:00
|
|
|
self.file.seek(SeekFrom::Start(csum_offset as u64))?;
|
2019-06-14 08:34:29 +00:00
|
|
|
self.file.write_all(&index_csum)?;
|
|
|
|
self.file.flush()?;
|
|
|
|
|
2018-12-16 12:39:21 +00:00
|
|
|
if let Err(err) = std::fs::rename(&self.tmp_filename, &self.filename) {
|
|
|
|
bail!("Atomic rename file {:?} failed - {}", self.filename, err);
|
|
|
|
}
|
|
|
|
|
2019-06-14 08:34:29 +00:00
|
|
|
Ok(index_csum)
|
2018-12-16 12:39:21 +00:00
|
|
|
}
|
|
|
|
|
2019-07-04 11:40:43 +00:00
|
|
|
pub fn check_chunk_alignment(&self, offset: usize, chunk_len: usize) -> Result<usize, Error> {
|
2019-06-13 09:47:23 +00:00
|
|
|
|
2019-07-04 11:40:43 +00:00
|
|
|
if offset < chunk_len {
|
|
|
|
bail!("got chunk with small offset ({} < {}", offset, chunk_len);
|
2019-06-13 09:47:23 +00:00
|
|
|
}
|
|
|
|
|
2019-07-04 11:40:43 +00:00
|
|
|
let pos = offset - chunk_len;
|
2018-12-15 13:51:05 +00:00
|
|
|
|
2019-07-04 11:40:43 +00:00
|
|
|
if offset > self.size {
|
|
|
|
bail!("chunk data exceeds size ({} >= {})", offset, self.size);
|
2018-12-15 13:51:05 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
// last chunk can be smaller
|
2019-07-04 11:40:43 +00:00
|
|
|
if ((offset != self.size) && (chunk_len != self.chunk_size)) ||
|
2019-06-13 09:47:23 +00:00
|
|
|
(chunk_len > self.chunk_size) || (chunk_len == 0) {
|
2019-07-04 11:40:43 +00:00
|
|
|
bail!("chunk with unexpected length ({} != {}", chunk_len, self.chunk_size);
|
2018-12-15 13:51:05 +00:00
|
|
|
}
|
|
|
|
|
2019-07-04 11:40:43 +00:00
|
|
|
if pos & (self.chunk_size-1) != 0 {
|
|
|
|
bail!("got unaligned chunk (pos = {})", pos);
|
2019-06-13 09:47:23 +00:00
|
|
|
}
|
|
|
|
|
2019-07-04 11:40:43 +00:00
|
|
|
Ok(pos / self.chunk_size)
|
|
|
|
}
|
|
|
|
|
|
|
|
// Note: We want to add data out of order, so do not assume any order here.
|
|
|
|
pub fn add_chunk(&mut self, chunk_info: &ChunkInfo, stat: &mut ChunkStat) -> Result<(), Error> {
|
|
|
|
|
|
|
|
let chunk_len = chunk_info.chunk_len as usize;
|
|
|
|
let offset = chunk_info.offset as usize; // end of chunk
|
|
|
|
|
|
|
|
let idx = self.check_chunk_alignment(offset, chunk_len)?;
|
|
|
|
|
2019-06-13 09:47:23 +00:00
|
|
|
let (is_duplicate, compressed_size) = self.store.insert_chunk(&chunk_info.chunk)?;
|
2019-02-25 10:36:05 +00:00
|
|
|
|
2019-05-29 05:08:34 +00:00
|
|
|
stat.chunk_count += 1;
|
|
|
|
stat.compressed_size += compressed_size;
|
2018-12-15 13:51:05 +00:00
|
|
|
|
2019-06-13 09:47:23 +00:00
|
|
|
let digest = chunk_info.chunk.digest();
|
|
|
|
|
2019-07-04 11:40:43 +00:00
|
|
|
println!("ADD CHUNK {} {} {}% {} {}", idx, chunk_len,
|
2019-06-14 09:40:04 +00:00
|
|
|
(compressed_size*100)/(chunk_len as u64), is_duplicate, proxmox::tools::digest_to_hex(digest));
|
2019-02-25 10:36:05 +00:00
|
|
|
|
|
|
|
if is_duplicate {
|
2019-05-29 05:08:34 +00:00
|
|
|
stat.duplicate_chunks += 1;
|
2019-02-25 10:36:05 +00:00
|
|
|
} else {
|
2019-05-29 05:08:34 +00:00
|
|
|
stat.disk_size += compressed_size;
|
2019-02-25 10:36:05 +00:00
|
|
|
}
|
2018-12-15 13:51:05 +00:00
|
|
|
|
2019-07-04 11:40:43 +00:00
|
|
|
self.add_digest(idx, digest)
|
2019-03-06 09:20:13 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
pub fn add_digest(&mut self, index: usize, digest: &[u8; 32]) -> Result<(), Error> {
|
2019-05-29 05:02:45 +00:00
|
|
|
|
2019-05-28 06:47:03 +00:00
|
|
|
if index >= self.index_length {
|
|
|
|
bail!("add digest failed - index out of range ({} >= {})", index, self.index_length);
|
|
|
|
}
|
|
|
|
|
2019-05-29 05:02:45 +00:00
|
|
|
if self.index == std::ptr::null_mut() { bail!("cannot write to closed index file."); }
|
|
|
|
|
2019-03-06 09:20:13 +00:00
|
|
|
let index_pos = index*32;
|
2018-12-15 13:51:05 +00:00
|
|
|
unsafe {
|
|
|
|
let dst = self.index.add(index_pos);
|
|
|
|
dst.copy_from_nonoverlapping(digest.as_ptr(), 32);
|
|
|
|
}
|
|
|
|
|
|
|
|
Ok(())
|
|
|
|
}
|
|
|
|
}
|
2019-07-05 07:19:56 +00:00
|
|
|
|
|
|
|
pub struct BufferedFixedReader<S> {
|
|
|
|
store: S,
|
|
|
|
index: FixedIndexReader,
|
|
|
|
archive_size: u64,
|
|
|
|
read_buffer: Vec<u8>,
|
|
|
|
buffered_chunk_idx: usize,
|
|
|
|
buffered_chunk_start: u64,
|
|
|
|
read_offset: u64,
|
|
|
|
}
|
|
|
|
|
|
|
|
impl <S: ReadChunk> BufferedFixedReader<S> {
|
|
|
|
|
|
|
|
pub fn new(index: FixedIndexReader, store: S) -> Self {
|
|
|
|
|
|
|
|
let archive_size = index.size;
|
|
|
|
Self {
|
|
|
|
store,
|
|
|
|
index: index,
|
|
|
|
archive_size: archive_size,
|
|
|
|
read_buffer: Vec::with_capacity(1024*1024),
|
|
|
|
buffered_chunk_idx: 0,
|
|
|
|
buffered_chunk_start: 0,
|
|
|
|
read_offset: 0,
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
pub fn archive_size(&self) -> u64 { self.archive_size }
|
|
|
|
|
|
|
|
fn buffer_chunk(&mut self, idx: usize) -> Result<(), Error> {
|
|
|
|
|
|
|
|
let index = &self.index;
|
|
|
|
let (start, end, digest) = index.chunk_info(idx)?;
|
|
|
|
|
|
|
|
// fixme: avoid copy
|
|
|
|
|
|
|
|
let data = self.store.read_chunk(&digest)?;
|
|
|
|
|
|
|
|
if (end - start) != data.len() as u64 {
|
|
|
|
bail!("read chunk with wrong size ({} != {}", (end - start), data.len());
|
|
|
|
}
|
|
|
|
|
|
|
|
self.read_buffer.clear();
|
|
|
|
self.read_buffer.extend_from_slice(&data);
|
|
|
|
|
|
|
|
self.buffered_chunk_idx = idx;
|
|
|
|
|
|
|
|
self.buffered_chunk_start = start as u64;
|
|
|
|
//println!("BUFFER {} {}", self.buffered_chunk_start, end);
|
|
|
|
Ok(())
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
impl <S: ReadChunk> crate::tools::BufferedRead for BufferedFixedReader<S> {
|
|
|
|
|
|
|
|
fn buffered_read(&mut self, offset: u64) -> Result<&[u8], Error> {
|
|
|
|
|
|
|
|
if offset == self.archive_size { return Ok(&self.read_buffer[0..0]); }
|
|
|
|
|
|
|
|
let buffer_len = self.read_buffer.len();
|
|
|
|
let index = &self.index;
|
|
|
|
|
|
|
|
// optimization for sequential read
|
|
|
|
if buffer_len > 0 &&
|
|
|
|
((self.buffered_chunk_idx + 1) < index.index_length) &&
|
|
|
|
(offset >= (self.buffered_chunk_start + (self.read_buffer.len() as u64)))
|
|
|
|
{
|
|
|
|
let next_idx = self.buffered_chunk_idx + 1;
|
|
|
|
let next_end = index.chunk_end(next_idx);
|
|
|
|
if offset < next_end {
|
|
|
|
self.buffer_chunk(next_idx)?;
|
|
|
|
let buffer_offset = (offset - self.buffered_chunk_start) as usize;
|
|
|
|
return Ok(&self.read_buffer[buffer_offset..]);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
if (buffer_len == 0) ||
|
|
|
|
(offset < self.buffered_chunk_start) ||
|
|
|
|
(offset >= (self.buffered_chunk_start + (self.read_buffer.len() as u64)))
|
|
|
|
{
|
|
|
|
let idx = (offset / index.chunk_size as u64) as usize;
|
|
|
|
self.buffer_chunk(idx)?;
|
|
|
|
}
|
|
|
|
|
|
|
|
let buffer_offset = (offset - self.buffered_chunk_start) as usize;
|
|
|
|
Ok(&self.read_buffer[buffer_offset..])
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
impl <S: ReadChunk> std::io::Read for BufferedFixedReader<S> {
|
|
|
|
|
|
|
|
fn read(&mut self, buf: &mut [u8]) -> Result<usize, std::io::Error> {
|
|
|
|
|
|
|
|
use std::io::{Error, ErrorKind};
|
|
|
|
use crate::tools::BufferedRead;
|
|
|
|
|
|
|
|
let data = match self.buffered_read(self.read_offset) {
|
|
|
|
Ok(v) => v,
|
|
|
|
Err(err) => return Err(Error::new(ErrorKind::Other, err.to_string())),
|
|
|
|
};
|
|
|
|
|
|
|
|
let n = if data.len() > buf.len() { buf.len() } else { data.len() };
|
|
|
|
|
|
|
|
unsafe { std::ptr::copy_nonoverlapping(data.as_ptr(), buf.as_mut_ptr(), n); }
|
|
|
|
|
|
|
|
self.read_offset += n as u64;
|
|
|
|
|
|
|
|
return Ok(n);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
impl <S: ReadChunk> Seek for BufferedFixedReader<S> {
|
|
|
|
|
|
|
|
fn seek(&mut self, pos: SeekFrom) -> Result<u64, std::io::Error> {
|
|
|
|
|
|
|
|
let new_offset = match pos {
|
|
|
|
SeekFrom::Start(start_offset) => start_offset as i64,
|
|
|
|
SeekFrom::End(end_offset) => (self.archive_size as i64)+ end_offset,
|
|
|
|
SeekFrom::Current(offset) => (self.read_offset as i64) + offset,
|
|
|
|
};
|
|
|
|
|
|
|
|
use std::io::{Error, ErrorKind};
|
|
|
|
if (new_offset < 0) || (new_offset > (self.archive_size as i64)) {
|
|
|
|
return Err(Error::new(
|
|
|
|
ErrorKind::Other,
|
|
|
|
format!("seek is out of range {} ([0..{}])", new_offset, self.archive_size)));
|
|
|
|
}
|
|
|
|
self.read_offset = new_offset as u64;
|
|
|
|
|
|
|
|
Ok(self.read_offset)
|
|
|
|
}
|
|
|
|
}
|