rust/hg-core/src/revlog/nodemap_docket.rs
author Raphaël Gomès <rgomes@octobus.net>
Thu, 03 Aug 2023 10:28:10 +0200
changeset 51198 51cc12158f97
parent 51120 532e74ad3ff6
permissions -rw-r--r--
rust-index: add `pack_header` support
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
46443
43d63979a75e rust: use HgError in RevlogError and Vfs
Simon Sapin <simon.sapin@octobus.net>
parents: 46390
diff changeset
     1
use crate::errors::{HgError, HgResultExt};
46390
0800aa42bb4c rust: use the bytes-cast crate to parse persistent nodemaps
Simon Sapin <simon.sapin@octobus.net>
parents: 46167
diff changeset
     2
use bytes_cast::{unaligned, BytesCast};
47955
e834b79def74 rust: Switch to the memmap2-rs crate
Simon Sapin <simon.sapin@octobus.net>
parents: 46630
diff changeset
     3
use memmap2::Mmap;
46090
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
     4
use std::path::{Path, PathBuf};
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
     5
49086
704e993e8ee9 rust-nodemap-docket: make unaware of `Repo`
Martin von Zweigbergk <martinvonz@google.com>
parents: 49085
diff changeset
     6
use crate::vfs::Vfs;
46090
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
     7
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
     8
const ONDISK_VERSION: u8 = 1;
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
     9
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    10
pub(super) struct NodeMapDocket {
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    11
    pub data_length: usize,
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    12
    // 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
    13
}
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    14
46390
0800aa42bb4c rust: use the bytes-cast crate to parse persistent nodemaps
Simon Sapin <simon.sapin@octobus.net>
parents: 46167
diff changeset
    15
#[derive(BytesCast)]
0800aa42bb4c rust: use the bytes-cast crate to parse persistent nodemaps
Simon Sapin <simon.sapin@octobus.net>
parents: 46167
diff changeset
    16
#[repr(C)]
0800aa42bb4c rust: use the bytes-cast crate to parse persistent nodemaps
Simon Sapin <simon.sapin@octobus.net>
parents: 46167
diff changeset
    17
struct DocketHeader {
0800aa42bb4c rust: use the bytes-cast crate to parse persistent nodemaps
Simon Sapin <simon.sapin@octobus.net>
parents: 46167
diff changeset
    18
    uid_size: u8,
0800aa42bb4c rust: use the bytes-cast crate to parse persistent nodemaps
Simon Sapin <simon.sapin@octobus.net>
parents: 46167
diff changeset
    19
    _tip_rev: unaligned::U64Be,
0800aa42bb4c rust: use the bytes-cast crate to parse persistent nodemaps
Simon Sapin <simon.sapin@octobus.net>
parents: 46167
diff changeset
    20
    data_length: unaligned::U64Be,
0800aa42bb4c rust: use the bytes-cast crate to parse persistent nodemaps
Simon Sapin <simon.sapin@octobus.net>
parents: 46167
diff changeset
    21
    _data_unused: unaligned::U64Be,
0800aa42bb4c rust: use the bytes-cast crate to parse persistent nodemaps
Simon Sapin <simon.sapin@octobus.net>
parents: 46167
diff changeset
    22
    tip_node_size: unaligned::U64Be,
0800aa42bb4c rust: use the bytes-cast crate to parse persistent nodemaps
Simon Sapin <simon.sapin@octobus.net>
parents: 46167
diff changeset
    23
}
0800aa42bb4c rust: use the bytes-cast crate to parse persistent nodemaps
Simon Sapin <simon.sapin@octobus.net>
parents: 46167
diff changeset
    24
46090
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    25
impl NodeMapDocket {
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    26
    /// 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
    27
    /// nodemap:
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    28
    ///
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    29
    /// * 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
    30
    ///   small revlogs), or
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    31
    /// * 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
    32
    ///   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
    33
    /// * 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
    34
    ///   can happen in a rare race condition).
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    35
    pub fn read_from_file(
49086
704e993e8ee9 rust-nodemap-docket: make unaware of `Repo`
Martin von Zweigbergk <martinvonz@google.com>
parents: 49085
diff changeset
    36
        store_vfs: &Vfs,
46090
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    37
        index_path: &Path,
47963
001d747c2baf rust: Return HgError instead of RevlogError in revlog constructors
Simon Sapin <simon.sapin@octobus.net>
parents: 47955
diff changeset
    38
    ) -> Result<Option<(Self, Mmap)>, HgError> {
46090
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    39
        let docket_path = index_path.with_extension("n");
46443
43d63979a75e rust: use HgError in RevlogError and Vfs
Simon Sapin <simon.sapin@octobus.net>
parents: 46390
diff changeset
    40
        let docket_bytes = if let Some(bytes) =
49086
704e993e8ee9 rust-nodemap-docket: make unaware of `Repo`
Martin von Zweigbergk <martinvonz@google.com>
parents: 49085
diff changeset
    41
            store_vfs.read(&docket_path).io_not_found_as_none()?
46443
43d63979a75e rust: use HgError in RevlogError and Vfs
Simon Sapin <simon.sapin@octobus.net>
parents: 46390
diff changeset
    42
        {
43d63979a75e rust: use HgError in RevlogError and Vfs
Simon Sapin <simon.sapin@octobus.net>
parents: 46390
diff changeset
    43
            bytes
43d63979a75e rust: use HgError in RevlogError and Vfs
Simon Sapin <simon.sapin@octobus.net>
parents: 46390
diff changeset
    44
        } else {
43d63979a75e rust: use HgError in RevlogError and Vfs
Simon Sapin <simon.sapin@octobus.net>
parents: 46390
diff changeset
    45
            return Ok(None);
46090
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    46
        };
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    47
46390
0800aa42bb4c rust: use the bytes-cast crate to parse persistent nodemaps
Simon Sapin <simon.sapin@octobus.net>
parents: 46167
diff changeset
    48
        let input = if let Some((&ONDISK_VERSION, rest)) =
46090
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    49
            docket_bytes.split_first()
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    50
        {
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    51
            rest
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    52
        } else {
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    53
            return Ok(None);
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    54
        };
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    55
46443
43d63979a75e rust: use HgError in RevlogError and Vfs
Simon Sapin <simon.sapin@octobus.net>
parents: 46390
diff changeset
    56
        /// Treat any error as a parse error
47963
001d747c2baf rust: Return HgError instead of RevlogError in revlog constructors
Simon Sapin <simon.sapin@octobus.net>
parents: 47955
diff changeset
    57
        fn parse<T, E>(result: Result<T, E>) -> Result<T, HgError> {
001d747c2baf rust: Return HgError instead of RevlogError in revlog constructors
Simon Sapin <simon.sapin@octobus.net>
parents: 47955
diff changeset
    58
            result
001d747c2baf rust: Return HgError instead of RevlogError in revlog constructors
Simon Sapin <simon.sapin@octobus.net>
parents: 47955
diff changeset
    59
                .map_err(|_| HgError::corrupted("nodemap docket parse error"))
46443
43d63979a75e rust: use HgError in RevlogError and Vfs
Simon Sapin <simon.sapin@octobus.net>
parents: 46390
diff changeset
    60
        }
43d63979a75e rust: use HgError in RevlogError and Vfs
Simon Sapin <simon.sapin@octobus.net>
parents: 46390
diff changeset
    61
43d63979a75e rust: use HgError in RevlogError and Vfs
Simon Sapin <simon.sapin@octobus.net>
parents: 46390
diff changeset
    62
        let (header, rest) = parse(DocketHeader::from_bytes(input))?;
46390
0800aa42bb4c rust: use the bytes-cast crate to parse persistent nodemaps
Simon Sapin <simon.sapin@octobus.net>
parents: 46167
diff changeset
    63
        let uid_size = header.uid_size as usize;
46090
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    64
        // 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
    65
        // systems?
46390
0800aa42bb4c rust: use the bytes-cast crate to parse persistent nodemaps
Simon Sapin <simon.sapin@octobus.net>
parents: 46167
diff changeset
    66
        let tip_node_size = header.tip_node_size.get() as usize;
0800aa42bb4c rust: use the bytes-cast crate to parse persistent nodemaps
Simon Sapin <simon.sapin@octobus.net>
parents: 46167
diff changeset
    67
        let data_length = header.data_length.get() as usize;
46443
43d63979a75e rust: use HgError in RevlogError and Vfs
Simon Sapin <simon.sapin@octobus.net>
parents: 46390
diff changeset
    68
        let (uid, rest) = parse(u8::slice_from_bytes(rest, uid_size))?;
43d63979a75e rust: use HgError in RevlogError and Vfs
Simon Sapin <simon.sapin@octobus.net>
parents: 46390
diff changeset
    69
        let (_tip_node, _rest) =
43d63979a75e rust: use HgError in RevlogError and Vfs
Simon Sapin <simon.sapin@octobus.net>
parents: 46390
diff changeset
    70
            parse(u8::slice_from_bytes(rest, tip_node_size))?;
43d63979a75e rust: use HgError in RevlogError and Vfs
Simon Sapin <simon.sapin@octobus.net>
parents: 46390
diff changeset
    71
        let uid = parse(std::str::from_utf8(uid))?;
46090
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    72
        let docket = NodeMapDocket { data_length };
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    73
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    74
        let data_path = rawdata_path(&docket_path, uid);
46443
43d63979a75e rust: use HgError in RevlogError and Vfs
Simon Sapin <simon.sapin@octobus.net>
parents: 46390
diff changeset
    75
        // TODO: use `vfs.read()` here when the `persistent-nodemap.mmap`
46090
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    76
        // config is false?
49086
704e993e8ee9 rust-nodemap-docket: make unaware of `Repo`
Martin von Zweigbergk <martinvonz@google.com>
parents: 49085
diff changeset
    77
        if let Some(mmap) =
51120
532e74ad3ff6 rust: run a clippy pass with the latest stable version
Raphaël Gomès <rgomes@octobus.net>
parents: 49632
diff changeset
    78
            store_vfs.mmap_open(data_path).io_not_found_as_none()?
46443
43d63979a75e rust: use HgError in RevlogError and Vfs
Simon Sapin <simon.sapin@octobus.net>
parents: 46390
diff changeset
    79
        {
43d63979a75e rust: use HgError in RevlogError and Vfs
Simon Sapin <simon.sapin@octobus.net>
parents: 46390
diff changeset
    80
            if mmap.len() >= data_length {
43d63979a75e rust: use HgError in RevlogError and Vfs
Simon Sapin <simon.sapin@octobus.net>
parents: 46390
diff changeset
    81
                Ok(Some((docket, mmap)))
43d63979a75e rust: use HgError in RevlogError and Vfs
Simon Sapin <simon.sapin@octobus.net>
parents: 46390
diff changeset
    82
            } else {
47963
001d747c2baf rust: Return HgError instead of RevlogError in revlog constructors
Simon Sapin <simon.sapin@octobus.net>
parents: 47955
diff changeset
    83
                Err(HgError::corrupted("persistent nodemap too short"))
46090
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    84
            }
46443
43d63979a75e rust: use HgError in RevlogError and Vfs
Simon Sapin <simon.sapin@octobus.net>
parents: 46390
diff changeset
    85
        } else {
46630
842f2372ced6 rhg: Don’t attempt to read persistent nodemap without .hg/requires opt-in
Simon Sapin <simon.sapin@octobus.net>
parents: 46443
diff changeset
    86
            // Even if .hg/requires opted in, some revlogs are deemed small
842f2372ced6 rhg: Don’t attempt to read persistent nodemap without .hg/requires opt-in
Simon Sapin <simon.sapin@octobus.net>
parents: 46443
diff changeset
    87
            // enough to not need a persistent nodemap.
46443
43d63979a75e rust: use HgError in RevlogError and Vfs
Simon Sapin <simon.sapin@octobus.net>
parents: 46390
diff changeset
    88
            Ok(None)
46090
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    89
        }
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    90
    }
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    91
}
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    92
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    93
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
    94
    let docket_name = docket_path
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    95
        .file_name()
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    96
        .expect("expected a base name")
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    97
        .to_str()
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
    98
        .expect("expected an ASCII file name in the store");
49632
048f829a445a hg-core: remove unneeded util now that we support Rust 1.42+
Raphaël Gomès <rgomes@octobus.net>
parents: 49086
diff changeset
    99
    let prefix = docket_name
048f829a445a hg-core: remove unneeded util now that we support Rust 1.42+
Raphaël Gomès <rgomes@octobus.net>
parents: 49086
diff changeset
   100
        .strip_suffix(".n.a")
048f829a445a hg-core: remove unneeded util now that we support Rust 1.42+
Raphaël Gomès <rgomes@octobus.net>
parents: 49086
diff changeset
   101
        .or_else(|| docket_name.strip_suffix(".n"))
46090
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
   102
        .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
   103
    let name = format!("{}-{}.nd", prefix, uid);
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
   104
    docket_path
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
   105
        .parent()
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
   106
        .expect("expected a non-root path")
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
   107
        .join(name)
9eb07ab3f2d4 rhg: use persistent nodemap when available
Simon Sapin <simon-commits@exyr.org>
parents:
diff changeset
   108
}