rust/hg-core/src/revlog/nodemap_docket.rs
author Simon Sapin <simon.sapin@octobus.net>
Mon, 14 Dec 2020 16:33:15 +0100
changeset 46167 8a4914397d02
parent 46090 9eb07ab3f2d4
child 46390 0800aa42bb4c
permissions -rw-r--r--
rust: introduce Repo and Vfs types for filesystem abstraction This is similar to the corresponding Python classes. Repo represents a repository and knows the path to the `.hg` directory, the `store` directory, and the working directory. Separating these will enable supporting the share extension. A Vfs is created from a Repo for one of these three directories. It has filesystem access APIs that take a relative std::path::Path as a parameter. Differential Revision: https://phab.mercurial-scm.org/D9596
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
46090
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
     1
use memmap::Mmap;
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
     2
use std::convert::TryInto;
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
     3
use std::path::{Path, PathBuf};
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
     4
46167
8a4914397d02 rust: introduce Repo and Vfs types for filesystem abstraction
Simon Sapin <simon.sapin@octobus.net>
parents: 46090
diff changeset
     5
use super::revlog::RevlogError;
8a4914397d02 rust: introduce Repo and Vfs types for filesystem abstraction
Simon Sapin <simon.sapin@octobus.net>
parents: 46090
diff changeset
     6
use crate::repo::Repo;
46090
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
     7
use crate::utils::strip_suffix;
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
     8
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
     9
const ONDISK_VERSION: u8 = 1;
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    10
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    11
pub(super) struct NodeMapDocket {
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    12
    pub data_length: usize,
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    13
    // TODO: keep here more of the data from `parse()` when we need it
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    14
}
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    15
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    16
impl NodeMapDocket {
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    17
    /// Return `Ok(None)` when the caller should proceed without a persistent
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    18
    /// nodemap:
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    19
    ///
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    20
    /// * This revlog does not have a `.n` docket file (it is not generated for
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    21
    ///   small revlogs), or
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    22
    /// * The docket has an unsupported version number (repositories created by
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    23
    ///   later hg, maybe that should be a requirement instead?), or
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    24
    /// * The docket file points to a missing (likely deleted) data file (this
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    25
    ///   can happen in a rare race condition).
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    26
    pub fn read_from_file(
46167
8a4914397d02 rust: introduce Repo and Vfs types for filesystem abstraction
Simon Sapin <simon.sapin@octobus.net>
parents: 46090
diff changeset
    27
        repo: &Repo,
46090
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    28
        index_path: &Path,
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    29
    ) -> Result<Option<(Self, Mmap)>, RevlogError> {
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    30
        let docket_path = index_path.with_extension("n");
46167
8a4914397d02 rust: introduce Repo and Vfs types for filesystem abstraction
Simon Sapin <simon.sapin@octobus.net>
parents: 46090
diff changeset
    31
        let docket_bytes = match repo.store_vfs().read(&docket_path) {
46090
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    32
            Err(e) if e.kind() == std::io::ErrorKind::NotFound => {
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    33
                return Ok(None)
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    34
            }
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    35
            Err(e) => return Err(RevlogError::IoError(e)),
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    36
            Ok(bytes) => bytes,
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    37
        };
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    38
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    39
        let mut input = if let Some((&ONDISK_VERSION, rest)) =
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    40
            docket_bytes.split_first()
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    41
        {
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    42
            rest
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    43
        } else {
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    44
            return Ok(None);
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    45
        };
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    46
        let input = &mut input;
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    47
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    48
        let uid_size = read_u8(input)? as usize;
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    49
        let _tip_rev = read_be_u64(input)?;
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    50
        // TODO: do we care about overflow for 4 GB+ nodemap files on 32-bit
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    51
        // systems?
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    52
        let data_length = read_be_u64(input)? as usize;
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    53
        let _data_unused = read_be_u64(input)?;
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    54
        let tip_node_size = read_be_u64(input)? as usize;
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    55
        let uid = read_bytes(input, uid_size)?;
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    56
        let _tip_node = read_bytes(input, tip_node_size)?;
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    57
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    58
        let uid =
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    59
            std::str::from_utf8(uid).map_err(|_| RevlogError::Corrupted)?;
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    60
        let docket = NodeMapDocket { data_length };
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    61
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    62
        let data_path = rawdata_path(&docket_path, uid);
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    63
        // TODO: use `std::fs::read` here when the `persistent-nodemap.mmap`
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    64
        // config is false?
46167
8a4914397d02 rust: introduce Repo and Vfs types for filesystem abstraction
Simon Sapin <simon.sapin@octobus.net>
parents: 46090
diff changeset
    65
        match repo.store_vfs().mmap_open(&data_path) {
46090
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    66
            Ok(mmap) => {
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    67
                if mmap.len() >= data_length {
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    68
                    Ok(Some((docket, mmap)))
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    69
                } else {
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    70
                    Err(RevlogError::Corrupted)
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    71
                }
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    72
            }
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    73
            Err(error) => {
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    74
                if error.kind() == std::io::ErrorKind::NotFound {
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    75
                    Ok(None)
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    76
                } else {
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    77
                    Err(RevlogError::IoError(error))
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    78
                }
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    79
            }
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    80
        }
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    81
    }
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    82
}
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    83
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    84
fn read_bytes<'a>(
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    85
    input: &mut &'a [u8],
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    86
    count: usize,
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    87
) -> Result<&'a [u8], RevlogError> {
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    88
    if let Some(start) = input.get(..count) {
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    89
        *input = &input[count..];
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    90
        Ok(start)
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    91
    } else {
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    92
        Err(RevlogError::Corrupted)
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    93
    }
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    94
}
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    95
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    96
fn read_u8<'a>(input: &mut &[u8]) -> Result<u8, RevlogError> {
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    97
    Ok(read_bytes(input, 1)?[0])
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    98
}
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    99
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
   100
fn read_be_u64<'a>(input: &mut &[u8]) -> Result<u64, RevlogError> {
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
   101
    let array = read_bytes(input, std::mem::size_of::<u64>())?
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
   102
        .try_into()
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
   103
        .unwrap();
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
   104
    Ok(u64::from_be_bytes(array))
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
   105
}
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
   106
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
   107
fn rawdata_path(docket_path: &Path, uid: &str) -> PathBuf {
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
   108
    let docket_name = docket_path
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
   109
        .file_name()
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
   110
        .expect("expected a base name")
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
   111
        .to_str()
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
   112
        .expect("expected an ASCII file name in the store");
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
   113
    let prefix = strip_suffix(docket_name, ".n.a")
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
   114
        .or_else(|| strip_suffix(docket_name, ".n"))
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
   115
        .expect("expected docket path in .n or .n.a");
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
   116
    let name = format!("{}-{}.nd", prefix, uid);
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
   117
    docket_path
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
   118
        .parent()
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
   119
        .expect("expected a non-root path")
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
   120
        .join(name)
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
   121
}