Show More
@@ -1,725 +1,727 b'' | |||||
1 | //! The "version 2" disk representation of the dirstate |
|
1 | //! The "version 2" disk representation of the dirstate | |
2 | //! |
|
2 | //! | |
3 | //! See `mercurial/helptext/internals/dirstate-v2.txt` |
|
3 | //! See `mercurial/helptext/internals/dirstate-v2.txt` | |
4 |
|
4 | |||
5 | use crate::dirstate::TruncatedTimestamp; |
|
5 | use crate::dirstate::TruncatedTimestamp; | |
6 | use crate::dirstate_tree::dirstate_map::{self, DirstateMap, NodeRef}; |
|
6 | use crate::dirstate_tree::dirstate_map::{self, DirstateMap, NodeRef}; | |
7 | use crate::dirstate_tree::path_with_basename::WithBasename; |
|
7 | use crate::dirstate_tree::path_with_basename::WithBasename; | |
8 | use crate::errors::HgError; |
|
8 | use crate::errors::HgError; | |
9 | use crate::utils::hg_path::HgPath; |
|
9 | use crate::utils::hg_path::HgPath; | |
10 | use crate::DirstateEntry; |
|
10 | use crate::DirstateEntry; | |
11 | use crate::DirstateError; |
|
11 | use crate::DirstateError; | |
12 | use crate::DirstateParents; |
|
12 | use crate::DirstateParents; | |
13 | use bitflags::bitflags; |
|
13 | use bitflags::bitflags; | |
14 | use bytes_cast::unaligned::{U16Be, U32Be}; |
|
14 | use bytes_cast::unaligned::{U16Be, U32Be}; | |
15 | use bytes_cast::BytesCast; |
|
15 | use bytes_cast::BytesCast; | |
16 | use format_bytes::format_bytes; |
|
16 | use format_bytes::format_bytes; | |
17 | use std::borrow::Cow; |
|
17 | use std::borrow::Cow; | |
18 | use std::convert::{TryFrom, TryInto}; |
|
18 | use std::convert::{TryFrom, TryInto}; | |
19 |
|
19 | |||
20 | /// Added at the start of `.hg/dirstate` when the "v2" format is used. |
|
20 | /// Added at the start of `.hg/dirstate` when the "v2" format is used. | |
21 | /// This a redundant sanity check more than an actual "magic number" since |
|
21 | /// This a redundant sanity check more than an actual "magic number" since | |
22 | /// `.hg/requires` already governs which format should be used. |
|
22 | /// `.hg/requires` already governs which format should be used. | |
23 | pub const V2_FORMAT_MARKER: &[u8; 12] = b"dirstate-v2\n"; |
|
23 | pub const V2_FORMAT_MARKER: &[u8; 12] = b"dirstate-v2\n"; | |
24 |
|
24 | |||
25 | /// Keep space for 256-bit hashes |
|
25 | /// Keep space for 256-bit hashes | |
26 | const STORED_NODE_ID_BYTES: usize = 32; |
|
26 | const STORED_NODE_ID_BYTES: usize = 32; | |
27 |
|
27 | |||
28 | /// β¦ even though only 160 bits are used for now, with SHA-1 |
|
28 | /// β¦ even though only 160 bits are used for now, with SHA-1 | |
29 | const USED_NODE_ID_BYTES: usize = 20; |
|
29 | const USED_NODE_ID_BYTES: usize = 20; | |
30 |
|
30 | |||
31 | pub(super) const IGNORE_PATTERNS_HASH_LEN: usize = 20; |
|
31 | pub(super) const IGNORE_PATTERNS_HASH_LEN: usize = 20; | |
32 | pub(super) type IgnorePatternsHash = [u8; IGNORE_PATTERNS_HASH_LEN]; |
|
32 | pub(super) type IgnorePatternsHash = [u8; IGNORE_PATTERNS_HASH_LEN]; | |
33 |
|
33 | |||
34 | /// Must match the constant of the same name in |
|
34 | /// Must match the constant of the same name in | |
35 | /// `mercurial/dirstateutils/docket.py` |
|
35 | /// `mercurial/dirstateutils/docket.py` | |
36 | const TREE_METADATA_SIZE: usize = 44; |
|
36 | const TREE_METADATA_SIZE: usize = 44; | |
37 |
|
37 | |||
|
38 | const NODE_SIZE: usize = 43; | |||
|
39 | ||||
38 | /// Make sure that size-affecting changes are made knowingly |
|
40 | /// Make sure that size-affecting changes are made knowingly | |
39 | #[allow(unused)] |
|
41 | #[allow(unused)] | |
40 | fn static_assert_size_of() { |
|
42 | fn static_assert_size_of() { | |
41 | let _ = std::mem::transmute::<TreeMetadata, [u8; TREE_METADATA_SIZE]>; |
|
43 | let _ = std::mem::transmute::<TreeMetadata, [u8; TREE_METADATA_SIZE]>; | |
42 | let _ = std::mem::transmute::<DocketHeader, [u8; TREE_METADATA_SIZE + 81]>; |
|
44 | let _ = std::mem::transmute::<DocketHeader, [u8; TREE_METADATA_SIZE + 81]>; | |
43 |
let _ = std::mem::transmute::<Node, [u8; |
|
45 | let _ = std::mem::transmute::<Node, [u8; NODE_SIZE]>; | |
44 | } |
|
46 | } | |
45 |
|
47 | |||
46 | // Must match `HEADER` in `mercurial/dirstateutils/docket.py` |
|
48 | // Must match `HEADER` in `mercurial/dirstateutils/docket.py` | |
47 | #[derive(BytesCast)] |
|
49 | #[derive(BytesCast)] | |
48 | #[repr(C)] |
|
50 | #[repr(C)] | |
49 | struct DocketHeader { |
|
51 | struct DocketHeader { | |
50 | marker: [u8; V2_FORMAT_MARKER.len()], |
|
52 | marker: [u8; V2_FORMAT_MARKER.len()], | |
51 | parent_1: [u8; STORED_NODE_ID_BYTES], |
|
53 | parent_1: [u8; STORED_NODE_ID_BYTES], | |
52 | parent_2: [u8; STORED_NODE_ID_BYTES], |
|
54 | parent_2: [u8; STORED_NODE_ID_BYTES], | |
53 |
|
55 | |||
54 | metadata: TreeMetadata, |
|
56 | metadata: TreeMetadata, | |
55 |
|
57 | |||
56 | /// Counted in bytes |
|
58 | /// Counted in bytes | |
57 | data_size: Size, |
|
59 | data_size: Size, | |
58 |
|
60 | |||
59 | uuid_size: u8, |
|
61 | uuid_size: u8, | |
60 | } |
|
62 | } | |
61 |
|
63 | |||
62 | pub struct Docket<'on_disk> { |
|
64 | pub struct Docket<'on_disk> { | |
63 | header: &'on_disk DocketHeader, |
|
65 | header: &'on_disk DocketHeader, | |
64 | uuid: &'on_disk [u8], |
|
66 | uuid: &'on_disk [u8], | |
65 | } |
|
67 | } | |
66 |
|
68 | |||
67 | /// Fields are documented in the *Tree metadata in the docket file* |
|
69 | /// Fields are documented in the *Tree metadata in the docket file* | |
68 | /// section of `mercurial/helptext/internals/dirstate-v2.txt` |
|
70 | /// section of `mercurial/helptext/internals/dirstate-v2.txt` | |
69 | #[derive(BytesCast)] |
|
71 | #[derive(BytesCast)] | |
70 | #[repr(C)] |
|
72 | #[repr(C)] | |
71 | struct TreeMetadata { |
|
73 | struct TreeMetadata { | |
72 | root_nodes: ChildNodes, |
|
74 | root_nodes: ChildNodes, | |
73 | nodes_with_entry_count: Size, |
|
75 | nodes_with_entry_count: Size, | |
74 | nodes_with_copy_source_count: Size, |
|
76 | nodes_with_copy_source_count: Size, | |
75 | unreachable_bytes: Size, |
|
77 | unreachable_bytes: Size, | |
76 | unused: [u8; 4], |
|
78 | unused: [u8; 4], | |
77 |
|
79 | |||
78 | /// See *Optional hash of ignore patterns* section of |
|
80 | /// See *Optional hash of ignore patterns* section of | |
79 | /// `mercurial/helptext/internals/dirstate-v2.txt` |
|
81 | /// `mercurial/helptext/internals/dirstate-v2.txt` | |
80 | ignore_patterns_hash: IgnorePatternsHash, |
|
82 | ignore_patterns_hash: IgnorePatternsHash, | |
81 | } |
|
83 | } | |
82 |
|
84 | |||
83 | /// Fields are documented in the *The data file format* |
|
85 | /// Fields are documented in the *The data file format* | |
84 | /// section of `mercurial/helptext/internals/dirstate-v2.txt` |
|
86 | /// section of `mercurial/helptext/internals/dirstate-v2.txt` | |
85 | #[derive(BytesCast)] |
|
87 | #[derive(BytesCast)] | |
86 | #[repr(C)] |
|
88 | #[repr(C)] | |
87 | pub(super) struct Node { |
|
89 | pub(super) struct Node { | |
88 | full_path: PathSlice, |
|
90 | full_path: PathSlice, | |
89 |
|
91 | |||
90 | /// In bytes from `self.full_path.start` |
|
92 | /// In bytes from `self.full_path.start` | |
91 | base_name_start: PathSize, |
|
93 | base_name_start: PathSize, | |
92 |
|
94 | |||
93 | copy_source: OptPathSlice, |
|
95 | copy_source: OptPathSlice, | |
94 | children: ChildNodes, |
|
96 | children: ChildNodes, | |
95 | pub(super) descendants_with_entry_count: Size, |
|
97 | pub(super) descendants_with_entry_count: Size, | |
96 | pub(super) tracked_descendants_count: Size, |
|
98 | pub(super) tracked_descendants_count: Size, | |
97 | flags: Flags, |
|
99 | flags: Flags, | |
98 | data: Entry, |
|
100 | data: Entry, | |
99 | } |
|
101 | } | |
100 |
|
102 | |||
101 | bitflags! { |
|
103 | bitflags! { | |
102 | #[derive(BytesCast)] |
|
104 | #[derive(BytesCast)] | |
103 | #[repr(C)] |
|
105 | #[repr(C)] | |
104 | struct Flags: u8 { |
|
106 | struct Flags: u8 { | |
105 | const WDIR_TRACKED = 1 << 0; |
|
107 | const WDIR_TRACKED = 1 << 0; | |
106 | const P1_TRACKED = 1 << 1; |
|
108 | const P1_TRACKED = 1 << 1; | |
107 | const P2_INFO = 1 << 2; |
|
109 | const P2_INFO = 1 << 2; | |
108 | const HAS_MODE_AND_SIZE = 1 << 3; |
|
110 | const HAS_MODE_AND_SIZE = 1 << 3; | |
109 | const HAS_MTIME = 1 << 4; |
|
111 | const HAS_MTIME = 1 << 4; | |
110 | const MODE_EXEC_PERM = 1 << 5; |
|
112 | const MODE_EXEC_PERM = 1 << 5; | |
111 | const MODE_IS_SYMLINK = 1 << 7; |
|
113 | const MODE_IS_SYMLINK = 1 << 7; | |
112 | } |
|
114 | } | |
113 | } |
|
115 | } | |
114 |
|
116 | |||
115 | #[derive(BytesCast, Copy, Clone, Debug)] |
|
117 | #[derive(BytesCast, Copy, Clone, Debug)] | |
116 | #[repr(C)] |
|
118 | #[repr(C)] | |
117 | struct Entry { |
|
119 | struct Entry { | |
118 | _padding: U32Be, |
|
120 | _padding: U32Be, | |
119 | size: U32Be, |
|
121 | size: U32Be, | |
120 | mtime: U32Be, |
|
122 | mtime: U32Be, | |
121 | } |
|
123 | } | |
122 |
|
124 | |||
123 | /// Duration since the Unix epoch |
|
125 | /// Duration since the Unix epoch | |
124 | #[derive(BytesCast, Copy, Clone)] |
|
126 | #[derive(BytesCast, Copy, Clone)] | |
125 | #[repr(C)] |
|
127 | #[repr(C)] | |
126 | struct PackedTimestamp { |
|
128 | struct PackedTimestamp { | |
127 | _padding: U32Be, |
|
129 | _padding: U32Be, | |
128 | truncated_seconds: U32Be, |
|
130 | truncated_seconds: U32Be, | |
129 | nanoseconds: U32Be, |
|
131 | nanoseconds: U32Be, | |
130 | } |
|
132 | } | |
131 |
|
133 | |||
132 | /// Counted in bytes from the start of the file |
|
134 | /// Counted in bytes from the start of the file | |
133 | /// |
|
135 | /// | |
134 | /// NOTE: not supporting `.hg/dirstate` files larger than 4 GiB. |
|
136 | /// NOTE: not supporting `.hg/dirstate` files larger than 4 GiB. | |
135 | type Offset = U32Be; |
|
137 | type Offset = U32Be; | |
136 |
|
138 | |||
137 | /// Counted in number of items |
|
139 | /// Counted in number of items | |
138 | /// |
|
140 | /// | |
139 | /// NOTE: we choose not to support counting more than 4 billion nodes anywhere. |
|
141 | /// NOTE: we choose not to support counting more than 4 billion nodes anywhere. | |
140 | type Size = U32Be; |
|
142 | type Size = U32Be; | |
141 |
|
143 | |||
142 | /// Counted in bytes |
|
144 | /// Counted in bytes | |
143 | /// |
|
145 | /// | |
144 | /// NOTE: we choose not to support file names/paths longer than 64 KiB. |
|
146 | /// NOTE: we choose not to support file names/paths longer than 64 KiB. | |
145 | type PathSize = U16Be; |
|
147 | type PathSize = U16Be; | |
146 |
|
148 | |||
147 | /// A contiguous sequence of `len` times `Node`, representing the child nodes |
|
149 | /// A contiguous sequence of `len` times `Node`, representing the child nodes | |
148 | /// of either some other node or of the repository root. |
|
150 | /// of either some other node or of the repository root. | |
149 | /// |
|
151 | /// | |
150 | /// Always sorted by ascending `full_path`, to allow binary search. |
|
152 | /// Always sorted by ascending `full_path`, to allow binary search. | |
151 | /// Since nodes with the same parent nodes also have the same parent path, |
|
153 | /// Since nodes with the same parent nodes also have the same parent path, | |
152 | /// only the `base_name`s need to be compared during binary search. |
|
154 | /// only the `base_name`s need to be compared during binary search. | |
153 | #[derive(BytesCast, Copy, Clone)] |
|
155 | #[derive(BytesCast, Copy, Clone)] | |
154 | #[repr(C)] |
|
156 | #[repr(C)] | |
155 | struct ChildNodes { |
|
157 | struct ChildNodes { | |
156 | start: Offset, |
|
158 | start: Offset, | |
157 | len: Size, |
|
159 | len: Size, | |
158 | } |
|
160 | } | |
159 |
|
161 | |||
160 | /// A `HgPath` of `len` bytes |
|
162 | /// A `HgPath` of `len` bytes | |
161 | #[derive(BytesCast, Copy, Clone)] |
|
163 | #[derive(BytesCast, Copy, Clone)] | |
162 | #[repr(C)] |
|
164 | #[repr(C)] | |
163 | struct PathSlice { |
|
165 | struct PathSlice { | |
164 | start: Offset, |
|
166 | start: Offset, | |
165 | len: PathSize, |
|
167 | len: PathSize, | |
166 | } |
|
168 | } | |
167 |
|
169 | |||
168 | /// Either nothing if `start == 0`, or a `HgPath` of `len` bytes |
|
170 | /// Either nothing if `start == 0`, or a `HgPath` of `len` bytes | |
169 | type OptPathSlice = PathSlice; |
|
171 | type OptPathSlice = PathSlice; | |
170 |
|
172 | |||
171 | /// Unexpected file format found in `.hg/dirstate` with the "v2" format. |
|
173 | /// Unexpected file format found in `.hg/dirstate` with the "v2" format. | |
172 | /// |
|
174 | /// | |
173 | /// This should only happen if Mercurial is buggy or a repository is corrupted. |
|
175 | /// This should only happen if Mercurial is buggy or a repository is corrupted. | |
174 | #[derive(Debug)] |
|
176 | #[derive(Debug)] | |
175 | pub struct DirstateV2ParseError; |
|
177 | pub struct DirstateV2ParseError; | |
176 |
|
178 | |||
177 | impl From<DirstateV2ParseError> for HgError { |
|
179 | impl From<DirstateV2ParseError> for HgError { | |
178 | fn from(_: DirstateV2ParseError) -> Self { |
|
180 | fn from(_: DirstateV2ParseError) -> Self { | |
179 | HgError::corrupted("dirstate-v2 parse error") |
|
181 | HgError::corrupted("dirstate-v2 parse error") | |
180 | } |
|
182 | } | |
181 | } |
|
183 | } | |
182 |
|
184 | |||
183 | impl From<DirstateV2ParseError> for crate::DirstateError { |
|
185 | impl From<DirstateV2ParseError> for crate::DirstateError { | |
184 | fn from(error: DirstateV2ParseError) -> Self { |
|
186 | fn from(error: DirstateV2ParseError) -> Self { | |
185 | HgError::from(error).into() |
|
187 | HgError::from(error).into() | |
186 | } |
|
188 | } | |
187 | } |
|
189 | } | |
188 |
|
190 | |||
189 | impl<'on_disk> Docket<'on_disk> { |
|
191 | impl<'on_disk> Docket<'on_disk> { | |
190 | pub fn parents(&self) -> DirstateParents { |
|
192 | pub fn parents(&self) -> DirstateParents { | |
191 | use crate::Node; |
|
193 | use crate::Node; | |
192 | let p1 = Node::try_from(&self.header.parent_1[..USED_NODE_ID_BYTES]) |
|
194 | let p1 = Node::try_from(&self.header.parent_1[..USED_NODE_ID_BYTES]) | |
193 | .unwrap() |
|
195 | .unwrap() | |
194 | .clone(); |
|
196 | .clone(); | |
195 | let p2 = Node::try_from(&self.header.parent_2[..USED_NODE_ID_BYTES]) |
|
197 | let p2 = Node::try_from(&self.header.parent_2[..USED_NODE_ID_BYTES]) | |
196 | .unwrap() |
|
198 | .unwrap() | |
197 | .clone(); |
|
199 | .clone(); | |
198 | DirstateParents { p1, p2 } |
|
200 | DirstateParents { p1, p2 } | |
199 | } |
|
201 | } | |
200 |
|
202 | |||
201 | pub fn tree_metadata(&self) -> &[u8] { |
|
203 | pub fn tree_metadata(&self) -> &[u8] { | |
202 | self.header.metadata.as_bytes() |
|
204 | self.header.metadata.as_bytes() | |
203 | } |
|
205 | } | |
204 |
|
206 | |||
205 | pub fn data_size(&self) -> usize { |
|
207 | pub fn data_size(&self) -> usize { | |
206 | // This `unwrap` could only panic on a 16-bit CPU |
|
208 | // This `unwrap` could only panic on a 16-bit CPU | |
207 | self.header.data_size.get().try_into().unwrap() |
|
209 | self.header.data_size.get().try_into().unwrap() | |
208 | } |
|
210 | } | |
209 |
|
211 | |||
210 | pub fn data_filename(&self) -> String { |
|
212 | pub fn data_filename(&self) -> String { | |
211 | String::from_utf8(format_bytes!(b"dirstate.{}", self.uuid)).unwrap() |
|
213 | String::from_utf8(format_bytes!(b"dirstate.{}", self.uuid)).unwrap() | |
212 | } |
|
214 | } | |
213 | } |
|
215 | } | |
214 |
|
216 | |||
215 | pub fn read_docket( |
|
217 | pub fn read_docket( | |
216 | on_disk: &[u8], |
|
218 | on_disk: &[u8], | |
217 | ) -> Result<Docket<'_>, DirstateV2ParseError> { |
|
219 | ) -> Result<Docket<'_>, DirstateV2ParseError> { | |
218 | let (header, uuid) = |
|
220 | let (header, uuid) = | |
219 | DocketHeader::from_bytes(on_disk).map_err(|_| DirstateV2ParseError)?; |
|
221 | DocketHeader::from_bytes(on_disk).map_err(|_| DirstateV2ParseError)?; | |
220 | let uuid_size = header.uuid_size as usize; |
|
222 | let uuid_size = header.uuid_size as usize; | |
221 | if header.marker == *V2_FORMAT_MARKER && uuid.len() == uuid_size { |
|
223 | if header.marker == *V2_FORMAT_MARKER && uuid.len() == uuid_size { | |
222 | Ok(Docket { header, uuid }) |
|
224 | Ok(Docket { header, uuid }) | |
223 | } else { |
|
225 | } else { | |
224 | Err(DirstateV2ParseError) |
|
226 | Err(DirstateV2ParseError) | |
225 | } |
|
227 | } | |
226 | } |
|
228 | } | |
227 |
|
229 | |||
228 | pub(super) fn read<'on_disk>( |
|
230 | pub(super) fn read<'on_disk>( | |
229 | on_disk: &'on_disk [u8], |
|
231 | on_disk: &'on_disk [u8], | |
230 | metadata: &[u8], |
|
232 | metadata: &[u8], | |
231 | ) -> Result<DirstateMap<'on_disk>, DirstateV2ParseError> { |
|
233 | ) -> Result<DirstateMap<'on_disk>, DirstateV2ParseError> { | |
232 | if on_disk.is_empty() { |
|
234 | if on_disk.is_empty() { | |
233 | return Ok(DirstateMap::empty(on_disk)); |
|
235 | return Ok(DirstateMap::empty(on_disk)); | |
234 | } |
|
236 | } | |
235 | let (meta, _) = TreeMetadata::from_bytes(metadata) |
|
237 | let (meta, _) = TreeMetadata::from_bytes(metadata) | |
236 | .map_err(|_| DirstateV2ParseError)?; |
|
238 | .map_err(|_| DirstateV2ParseError)?; | |
237 | let dirstate_map = DirstateMap { |
|
239 | let dirstate_map = DirstateMap { | |
238 | on_disk, |
|
240 | on_disk, | |
239 | root: dirstate_map::ChildNodes::OnDisk(read_nodes( |
|
241 | root: dirstate_map::ChildNodes::OnDisk(read_nodes( | |
240 | on_disk, |
|
242 | on_disk, | |
241 | meta.root_nodes, |
|
243 | meta.root_nodes, | |
242 | )?), |
|
244 | )?), | |
243 | nodes_with_entry_count: meta.nodes_with_entry_count.get(), |
|
245 | nodes_with_entry_count: meta.nodes_with_entry_count.get(), | |
244 | nodes_with_copy_source_count: meta.nodes_with_copy_source_count.get(), |
|
246 | nodes_with_copy_source_count: meta.nodes_with_copy_source_count.get(), | |
245 | ignore_patterns_hash: meta.ignore_patterns_hash, |
|
247 | ignore_patterns_hash: meta.ignore_patterns_hash, | |
246 | unreachable_bytes: meta.unreachable_bytes.get(), |
|
248 | unreachable_bytes: meta.unreachable_bytes.get(), | |
247 | }; |
|
249 | }; | |
248 | Ok(dirstate_map) |
|
250 | Ok(dirstate_map) | |
249 | } |
|
251 | } | |
250 |
|
252 | |||
251 | impl Node { |
|
253 | impl Node { | |
252 | pub(super) fn full_path<'on_disk>( |
|
254 | pub(super) fn full_path<'on_disk>( | |
253 | &self, |
|
255 | &self, | |
254 | on_disk: &'on_disk [u8], |
|
256 | on_disk: &'on_disk [u8], | |
255 | ) -> Result<&'on_disk HgPath, DirstateV2ParseError> { |
|
257 | ) -> Result<&'on_disk HgPath, DirstateV2ParseError> { | |
256 | read_hg_path(on_disk, self.full_path) |
|
258 | read_hg_path(on_disk, self.full_path) | |
257 | } |
|
259 | } | |
258 |
|
260 | |||
259 | pub(super) fn base_name_start<'on_disk>( |
|
261 | pub(super) fn base_name_start<'on_disk>( | |
260 | &self, |
|
262 | &self, | |
261 | ) -> Result<usize, DirstateV2ParseError> { |
|
263 | ) -> Result<usize, DirstateV2ParseError> { | |
262 | let start = self.base_name_start.get(); |
|
264 | let start = self.base_name_start.get(); | |
263 | if start < self.full_path.len.get() { |
|
265 | if start < self.full_path.len.get() { | |
264 | let start = usize::try_from(start) |
|
266 | let start = usize::try_from(start) | |
265 | // u32 -> usize, could only panic on a 16-bit CPU |
|
267 | // u32 -> usize, could only panic on a 16-bit CPU | |
266 | .expect("dirstate-v2 base_name_start out of bounds"); |
|
268 | .expect("dirstate-v2 base_name_start out of bounds"); | |
267 | Ok(start) |
|
269 | Ok(start) | |
268 | } else { |
|
270 | } else { | |
269 | Err(DirstateV2ParseError) |
|
271 | Err(DirstateV2ParseError) | |
270 | } |
|
272 | } | |
271 | } |
|
273 | } | |
272 |
|
274 | |||
273 | pub(super) fn base_name<'on_disk>( |
|
275 | pub(super) fn base_name<'on_disk>( | |
274 | &self, |
|
276 | &self, | |
275 | on_disk: &'on_disk [u8], |
|
277 | on_disk: &'on_disk [u8], | |
276 | ) -> Result<&'on_disk HgPath, DirstateV2ParseError> { |
|
278 | ) -> Result<&'on_disk HgPath, DirstateV2ParseError> { | |
277 | let full_path = self.full_path(on_disk)?; |
|
279 | let full_path = self.full_path(on_disk)?; | |
278 | let base_name_start = self.base_name_start()?; |
|
280 | let base_name_start = self.base_name_start()?; | |
279 | Ok(HgPath::new(&full_path.as_bytes()[base_name_start..])) |
|
281 | Ok(HgPath::new(&full_path.as_bytes()[base_name_start..])) | |
280 | } |
|
282 | } | |
281 |
|
283 | |||
282 | pub(super) fn path<'on_disk>( |
|
284 | pub(super) fn path<'on_disk>( | |
283 | &self, |
|
285 | &self, | |
284 | on_disk: &'on_disk [u8], |
|
286 | on_disk: &'on_disk [u8], | |
285 | ) -> Result<dirstate_map::NodeKey<'on_disk>, DirstateV2ParseError> { |
|
287 | ) -> Result<dirstate_map::NodeKey<'on_disk>, DirstateV2ParseError> { | |
286 | Ok(WithBasename::from_raw_parts( |
|
288 | Ok(WithBasename::from_raw_parts( | |
287 | Cow::Borrowed(self.full_path(on_disk)?), |
|
289 | Cow::Borrowed(self.full_path(on_disk)?), | |
288 | self.base_name_start()?, |
|
290 | self.base_name_start()?, | |
289 | )) |
|
291 | )) | |
290 | } |
|
292 | } | |
291 |
|
293 | |||
292 | pub(super) fn has_copy_source<'on_disk>(&self) -> bool { |
|
294 | pub(super) fn has_copy_source<'on_disk>(&self) -> bool { | |
293 | self.copy_source.start.get() != 0 |
|
295 | self.copy_source.start.get() != 0 | |
294 | } |
|
296 | } | |
295 |
|
297 | |||
296 | pub(super) fn copy_source<'on_disk>( |
|
298 | pub(super) fn copy_source<'on_disk>( | |
297 | &self, |
|
299 | &self, | |
298 | on_disk: &'on_disk [u8], |
|
300 | on_disk: &'on_disk [u8], | |
299 | ) -> Result<Option<&'on_disk HgPath>, DirstateV2ParseError> { |
|
301 | ) -> Result<Option<&'on_disk HgPath>, DirstateV2ParseError> { | |
300 | Ok(if self.has_copy_source() { |
|
302 | Ok(if self.has_copy_source() { | |
301 | Some(read_hg_path(on_disk, self.copy_source)?) |
|
303 | Some(read_hg_path(on_disk, self.copy_source)?) | |
302 | } else { |
|
304 | } else { | |
303 | None |
|
305 | None | |
304 | }) |
|
306 | }) | |
305 | } |
|
307 | } | |
306 |
|
308 | |||
307 | fn has_entry(&self) -> bool { |
|
309 | fn has_entry(&self) -> bool { | |
308 | self.flags.intersects( |
|
310 | self.flags.intersects( | |
309 | Flags::WDIR_TRACKED | Flags::P1_TRACKED | Flags::P2_INFO, |
|
311 | Flags::WDIR_TRACKED | Flags::P1_TRACKED | Flags::P2_INFO, | |
310 | ) |
|
312 | ) | |
311 | } |
|
313 | } | |
312 |
|
314 | |||
313 | pub(super) fn node_data( |
|
315 | pub(super) fn node_data( | |
314 | &self, |
|
316 | &self, | |
315 | ) -> Result<dirstate_map::NodeData, DirstateV2ParseError> { |
|
317 | ) -> Result<dirstate_map::NodeData, DirstateV2ParseError> { | |
316 | if self.has_entry() { |
|
318 | if self.has_entry() { | |
317 | Ok(dirstate_map::NodeData::Entry(self.assume_entry())) |
|
319 | Ok(dirstate_map::NodeData::Entry(self.assume_entry())) | |
318 | } else if let Some(mtime) = self.cached_directory_mtime()? { |
|
320 | } else if let Some(mtime) = self.cached_directory_mtime()? { | |
319 | Ok(dirstate_map::NodeData::CachedDirectory { mtime }) |
|
321 | Ok(dirstate_map::NodeData::CachedDirectory { mtime }) | |
320 | } else { |
|
322 | } else { | |
321 | Ok(dirstate_map::NodeData::None) |
|
323 | Ok(dirstate_map::NodeData::None) | |
322 | } |
|
324 | } | |
323 | } |
|
325 | } | |
324 |
|
326 | |||
325 | pub(super) fn cached_directory_mtime( |
|
327 | pub(super) fn cached_directory_mtime( | |
326 | &self, |
|
328 | &self, | |
327 | ) -> Result<Option<TruncatedTimestamp>, DirstateV2ParseError> { |
|
329 | ) -> Result<Option<TruncatedTimestamp>, DirstateV2ParseError> { | |
328 | Ok( |
|
330 | Ok( | |
329 | if self.flags.contains(Flags::HAS_MTIME) && !self.has_entry() { |
|
331 | if self.flags.contains(Flags::HAS_MTIME) && !self.has_entry() { | |
330 | Some(self.data.as_timestamp()?) |
|
332 | Some(self.data.as_timestamp()?) | |
331 | } else { |
|
333 | } else { | |
332 | None |
|
334 | None | |
333 | }, |
|
335 | }, | |
334 | ) |
|
336 | ) | |
335 | } |
|
337 | } | |
336 |
|
338 | |||
337 | fn synthesize_unix_mode(&self) -> u32 { |
|
339 | fn synthesize_unix_mode(&self) -> u32 { | |
338 | let file_type = if self.flags.contains(Flags::MODE_IS_SYMLINK) { |
|
340 | let file_type = if self.flags.contains(Flags::MODE_IS_SYMLINK) { | |
339 | libc::S_IFLNK |
|
341 | libc::S_IFLNK | |
340 | } else { |
|
342 | } else { | |
341 | libc::S_IFREG |
|
343 | libc::S_IFREG | |
342 | }; |
|
344 | }; | |
343 | let permisions = if self.flags.contains(Flags::MODE_EXEC_PERM) { |
|
345 | let permisions = if self.flags.contains(Flags::MODE_EXEC_PERM) { | |
344 | 0o755 |
|
346 | 0o755 | |
345 | } else { |
|
347 | } else { | |
346 | 0o644 |
|
348 | 0o644 | |
347 | }; |
|
349 | }; | |
348 | file_type | permisions |
|
350 | file_type | permisions | |
349 | } |
|
351 | } | |
350 |
|
352 | |||
351 | fn assume_entry(&self) -> DirstateEntry { |
|
353 | fn assume_entry(&self) -> DirstateEntry { | |
352 | // TODO: convert through raw bits instead? |
|
354 | // TODO: convert through raw bits instead? | |
353 | let wdir_tracked = self.flags.contains(Flags::WDIR_TRACKED); |
|
355 | let wdir_tracked = self.flags.contains(Flags::WDIR_TRACKED); | |
354 | let p1_tracked = self.flags.contains(Flags::P1_TRACKED); |
|
356 | let p1_tracked = self.flags.contains(Flags::P1_TRACKED); | |
355 | let p2_info = self.flags.contains(Flags::P2_INFO); |
|
357 | let p2_info = self.flags.contains(Flags::P2_INFO); | |
356 | let mode_size = if self.flags.contains(Flags::HAS_MODE_AND_SIZE) { |
|
358 | let mode_size = if self.flags.contains(Flags::HAS_MODE_AND_SIZE) { | |
357 | Some((self.synthesize_unix_mode(), self.data.size.into())) |
|
359 | Some((self.synthesize_unix_mode(), self.data.size.into())) | |
358 | } else { |
|
360 | } else { | |
359 | None |
|
361 | None | |
360 | }; |
|
362 | }; | |
361 | let mtime = if self.flags.contains(Flags::HAS_MTIME) { |
|
363 | let mtime = if self.flags.contains(Flags::HAS_MTIME) { | |
362 | Some(self.data.mtime.into()) |
|
364 | Some(self.data.mtime.into()) | |
363 | } else { |
|
365 | } else { | |
364 | None |
|
366 | None | |
365 | }; |
|
367 | }; | |
366 | DirstateEntry::from_v2_data( |
|
368 | DirstateEntry::from_v2_data( | |
367 | wdir_tracked, |
|
369 | wdir_tracked, | |
368 | p1_tracked, |
|
370 | p1_tracked, | |
369 | p2_info, |
|
371 | p2_info, | |
370 | mode_size, |
|
372 | mode_size, | |
371 | mtime, |
|
373 | mtime, | |
372 | ) |
|
374 | ) | |
373 | } |
|
375 | } | |
374 |
|
376 | |||
375 | pub(super) fn entry( |
|
377 | pub(super) fn entry( | |
376 | &self, |
|
378 | &self, | |
377 | ) -> Result<Option<DirstateEntry>, DirstateV2ParseError> { |
|
379 | ) -> Result<Option<DirstateEntry>, DirstateV2ParseError> { | |
378 | if self.has_entry() { |
|
380 | if self.has_entry() { | |
379 | Ok(Some(self.assume_entry())) |
|
381 | Ok(Some(self.assume_entry())) | |
380 | } else { |
|
382 | } else { | |
381 | Ok(None) |
|
383 | Ok(None) | |
382 | } |
|
384 | } | |
383 | } |
|
385 | } | |
384 |
|
386 | |||
385 | pub(super) fn children<'on_disk>( |
|
387 | pub(super) fn children<'on_disk>( | |
386 | &self, |
|
388 | &self, | |
387 | on_disk: &'on_disk [u8], |
|
389 | on_disk: &'on_disk [u8], | |
388 | ) -> Result<&'on_disk [Node], DirstateV2ParseError> { |
|
390 | ) -> Result<&'on_disk [Node], DirstateV2ParseError> { | |
389 | read_nodes(on_disk, self.children) |
|
391 | read_nodes(on_disk, self.children) | |
390 | } |
|
392 | } | |
391 |
|
393 | |||
392 | pub(super) fn to_in_memory_node<'on_disk>( |
|
394 | pub(super) fn to_in_memory_node<'on_disk>( | |
393 | &self, |
|
395 | &self, | |
394 | on_disk: &'on_disk [u8], |
|
396 | on_disk: &'on_disk [u8], | |
395 | ) -> Result<dirstate_map::Node<'on_disk>, DirstateV2ParseError> { |
|
397 | ) -> Result<dirstate_map::Node<'on_disk>, DirstateV2ParseError> { | |
396 | Ok(dirstate_map::Node { |
|
398 | Ok(dirstate_map::Node { | |
397 | children: dirstate_map::ChildNodes::OnDisk( |
|
399 | children: dirstate_map::ChildNodes::OnDisk( | |
398 | self.children(on_disk)?, |
|
400 | self.children(on_disk)?, | |
399 | ), |
|
401 | ), | |
400 | copy_source: self.copy_source(on_disk)?.map(Cow::Borrowed), |
|
402 | copy_source: self.copy_source(on_disk)?.map(Cow::Borrowed), | |
401 | data: self.node_data()?, |
|
403 | data: self.node_data()?, | |
402 | descendants_with_entry_count: self |
|
404 | descendants_with_entry_count: self | |
403 | .descendants_with_entry_count |
|
405 | .descendants_with_entry_count | |
404 | .get(), |
|
406 | .get(), | |
405 | tracked_descendants_count: self.tracked_descendants_count.get(), |
|
407 | tracked_descendants_count: self.tracked_descendants_count.get(), | |
406 | }) |
|
408 | }) | |
407 | } |
|
409 | } | |
408 | } |
|
410 | } | |
409 |
|
411 | |||
410 | impl Entry { |
|
412 | impl Entry { | |
411 | fn from_dirstate_entry(entry: &DirstateEntry) -> (Flags, Self) { |
|
413 | fn from_dirstate_entry(entry: &DirstateEntry) -> (Flags, Self) { | |
412 | let (wdir_tracked, p1_tracked, p2_info, mode_size_opt, mtime_opt) = |
|
414 | let (wdir_tracked, p1_tracked, p2_info, mode_size_opt, mtime_opt) = | |
413 | entry.v2_data(); |
|
415 | entry.v2_data(); | |
414 | // TODO: convert throug raw flag bits instead? |
|
416 | // TODO: convert throug raw flag bits instead? | |
415 | let mut flags = Flags::empty(); |
|
417 | let mut flags = Flags::empty(); | |
416 | flags.set(Flags::WDIR_TRACKED, wdir_tracked); |
|
418 | flags.set(Flags::WDIR_TRACKED, wdir_tracked); | |
417 | flags.set(Flags::P1_TRACKED, p1_tracked); |
|
419 | flags.set(Flags::P1_TRACKED, p1_tracked); | |
418 | flags.set(Flags::P2_INFO, p2_info); |
|
420 | flags.set(Flags::P2_INFO, p2_info); | |
419 | let (size, mtime); |
|
421 | let (size, mtime); | |
420 | if let Some((m, s)) = mode_size_opt { |
|
422 | if let Some((m, s)) = mode_size_opt { | |
421 | let exec_perm = m & libc::S_IXUSR != 0; |
|
423 | let exec_perm = m & libc::S_IXUSR != 0; | |
422 | let is_symlink = m & libc::S_IFMT == libc::S_IFLNK; |
|
424 | let is_symlink = m & libc::S_IFMT == libc::S_IFLNK; | |
423 | flags.set(Flags::MODE_EXEC_PERM, exec_perm); |
|
425 | flags.set(Flags::MODE_EXEC_PERM, exec_perm); | |
424 | flags.set(Flags::MODE_IS_SYMLINK, is_symlink); |
|
426 | flags.set(Flags::MODE_IS_SYMLINK, is_symlink); | |
425 | size = s; |
|
427 | size = s; | |
426 | flags.insert(Flags::HAS_MODE_AND_SIZE) |
|
428 | flags.insert(Flags::HAS_MODE_AND_SIZE) | |
427 | } else { |
|
429 | } else { | |
428 | size = 0; |
|
430 | size = 0; | |
429 | } |
|
431 | } | |
430 | if let Some(m) = mtime_opt { |
|
432 | if let Some(m) = mtime_opt { | |
431 | mtime = m; |
|
433 | mtime = m; | |
432 | flags.insert(Flags::HAS_MTIME); |
|
434 | flags.insert(Flags::HAS_MTIME); | |
433 | } else { |
|
435 | } else { | |
434 | mtime = 0; |
|
436 | mtime = 0; | |
435 | } |
|
437 | } | |
436 | let raw_entry = Entry { |
|
438 | let raw_entry = Entry { | |
437 | _padding: 0.into(), |
|
439 | _padding: 0.into(), | |
438 | size: size.into(), |
|
440 | size: size.into(), | |
439 | mtime: mtime.into(), |
|
441 | mtime: mtime.into(), | |
440 | }; |
|
442 | }; | |
441 | (flags, raw_entry) |
|
443 | (flags, raw_entry) | |
442 | } |
|
444 | } | |
443 |
|
445 | |||
444 | fn from_timestamp(timestamp: TruncatedTimestamp) -> Self { |
|
446 | fn from_timestamp(timestamp: TruncatedTimestamp) -> Self { | |
445 | let packed = PackedTimestamp { |
|
447 | let packed = PackedTimestamp { | |
446 | _padding: 0.into(), |
|
448 | _padding: 0.into(), | |
447 | truncated_seconds: timestamp.truncated_seconds().into(), |
|
449 | truncated_seconds: timestamp.truncated_seconds().into(), | |
448 | nanoseconds: timestamp.nanoseconds().into(), |
|
450 | nanoseconds: timestamp.nanoseconds().into(), | |
449 | }; |
|
451 | }; | |
450 | // Safety: both types implement the `ByteCast` trait, so we could |
|
452 | // Safety: both types implement the `ByteCast` trait, so we could | |
451 | // safely use `as_bytes` and `from_bytes` to do this conversion. Using |
|
453 | // safely use `as_bytes` and `from_bytes` to do this conversion. Using | |
452 | // `transmute` instead makes the compiler check that the two types |
|
454 | // `transmute` instead makes the compiler check that the two types | |
453 | // have the same size, which eliminates the error case of |
|
455 | // have the same size, which eliminates the error case of | |
454 | // `from_bytes`. |
|
456 | // `from_bytes`. | |
455 | unsafe { std::mem::transmute::<PackedTimestamp, Entry>(packed) } |
|
457 | unsafe { std::mem::transmute::<PackedTimestamp, Entry>(packed) } | |
456 | } |
|
458 | } | |
457 |
|
459 | |||
458 | fn as_timestamp(self) -> Result<TruncatedTimestamp, DirstateV2ParseError> { |
|
460 | fn as_timestamp(self) -> Result<TruncatedTimestamp, DirstateV2ParseError> { | |
459 | // Safety: same as above in `from_timestamp` |
|
461 | // Safety: same as above in `from_timestamp` | |
460 | let packed = |
|
462 | let packed = | |
461 | unsafe { std::mem::transmute::<Entry, PackedTimestamp>(self) }; |
|
463 | unsafe { std::mem::transmute::<Entry, PackedTimestamp>(self) }; | |
462 | TruncatedTimestamp::from_already_truncated( |
|
464 | TruncatedTimestamp::from_already_truncated( | |
463 | packed.truncated_seconds.get(), |
|
465 | packed.truncated_seconds.get(), | |
464 | packed.nanoseconds.get(), |
|
466 | packed.nanoseconds.get(), | |
465 | ) |
|
467 | ) | |
466 | } |
|
468 | } | |
467 | } |
|
469 | } | |
468 |
|
470 | |||
469 | fn read_hg_path( |
|
471 | fn read_hg_path( | |
470 | on_disk: &[u8], |
|
472 | on_disk: &[u8], | |
471 | slice: PathSlice, |
|
473 | slice: PathSlice, | |
472 | ) -> Result<&HgPath, DirstateV2ParseError> { |
|
474 | ) -> Result<&HgPath, DirstateV2ParseError> { | |
473 | read_slice(on_disk, slice.start, slice.len.get()).map(HgPath::new) |
|
475 | read_slice(on_disk, slice.start, slice.len.get()).map(HgPath::new) | |
474 | } |
|
476 | } | |
475 |
|
477 | |||
476 | fn read_nodes( |
|
478 | fn read_nodes( | |
477 | on_disk: &[u8], |
|
479 | on_disk: &[u8], | |
478 | slice: ChildNodes, |
|
480 | slice: ChildNodes, | |
479 | ) -> Result<&[Node], DirstateV2ParseError> { |
|
481 | ) -> Result<&[Node], DirstateV2ParseError> { | |
480 | read_slice(on_disk, slice.start, slice.len.get()) |
|
482 | read_slice(on_disk, slice.start, slice.len.get()) | |
481 | } |
|
483 | } | |
482 |
|
484 | |||
483 | fn read_slice<T, Len>( |
|
485 | fn read_slice<T, Len>( | |
484 | on_disk: &[u8], |
|
486 | on_disk: &[u8], | |
485 | start: Offset, |
|
487 | start: Offset, | |
486 | len: Len, |
|
488 | len: Len, | |
487 | ) -> Result<&[T], DirstateV2ParseError> |
|
489 | ) -> Result<&[T], DirstateV2ParseError> | |
488 | where |
|
490 | where | |
489 | T: BytesCast, |
|
491 | T: BytesCast, | |
490 | Len: TryInto<usize>, |
|
492 | Len: TryInto<usize>, | |
491 | { |
|
493 | { | |
492 | // Either `usize::MAX` would result in "out of bounds" error since a single |
|
494 | // Either `usize::MAX` would result in "out of bounds" error since a single | |
493 | // `&[u8]` cannot occupy the entire addess space. |
|
495 | // `&[u8]` cannot occupy the entire addess space. | |
494 | let start = start.get().try_into().unwrap_or(std::usize::MAX); |
|
496 | let start = start.get().try_into().unwrap_or(std::usize::MAX); | |
495 | let len = len.try_into().unwrap_or(std::usize::MAX); |
|
497 | let len = len.try_into().unwrap_or(std::usize::MAX); | |
496 | on_disk |
|
498 | on_disk | |
497 | .get(start..) |
|
499 | .get(start..) | |
498 | .and_then(|bytes| T::slice_from_bytes(bytes, len).ok()) |
|
500 | .and_then(|bytes| T::slice_from_bytes(bytes, len).ok()) | |
499 | .map(|(slice, _rest)| slice) |
|
501 | .map(|(slice, _rest)| slice) | |
500 | .ok_or_else(|| DirstateV2ParseError) |
|
502 | .ok_or_else(|| DirstateV2ParseError) | |
501 | } |
|
503 | } | |
502 |
|
504 | |||
503 | pub(crate) fn for_each_tracked_path<'on_disk>( |
|
505 | pub(crate) fn for_each_tracked_path<'on_disk>( | |
504 | on_disk: &'on_disk [u8], |
|
506 | on_disk: &'on_disk [u8], | |
505 | metadata: &[u8], |
|
507 | metadata: &[u8], | |
506 | mut f: impl FnMut(&'on_disk HgPath), |
|
508 | mut f: impl FnMut(&'on_disk HgPath), | |
507 | ) -> Result<(), DirstateV2ParseError> { |
|
509 | ) -> Result<(), DirstateV2ParseError> { | |
508 | let (meta, _) = TreeMetadata::from_bytes(metadata) |
|
510 | let (meta, _) = TreeMetadata::from_bytes(metadata) | |
509 | .map_err(|_| DirstateV2ParseError)?; |
|
511 | .map_err(|_| DirstateV2ParseError)?; | |
510 | fn recur<'on_disk>( |
|
512 | fn recur<'on_disk>( | |
511 | on_disk: &'on_disk [u8], |
|
513 | on_disk: &'on_disk [u8], | |
512 | nodes: ChildNodes, |
|
514 | nodes: ChildNodes, | |
513 | f: &mut impl FnMut(&'on_disk HgPath), |
|
515 | f: &mut impl FnMut(&'on_disk HgPath), | |
514 | ) -> Result<(), DirstateV2ParseError> { |
|
516 | ) -> Result<(), DirstateV2ParseError> { | |
515 | for node in read_nodes(on_disk, nodes)? { |
|
517 | for node in read_nodes(on_disk, nodes)? { | |
516 | if let Some(entry) = node.entry()? { |
|
518 | if let Some(entry) = node.entry()? { | |
517 | if entry.state().is_tracked() { |
|
519 | if entry.state().is_tracked() { | |
518 | f(node.full_path(on_disk)?) |
|
520 | f(node.full_path(on_disk)?) | |
519 | } |
|
521 | } | |
520 | } |
|
522 | } | |
521 | recur(on_disk, node.children, f)? |
|
523 | recur(on_disk, node.children, f)? | |
522 | } |
|
524 | } | |
523 | Ok(()) |
|
525 | Ok(()) | |
524 | } |
|
526 | } | |
525 | recur(on_disk, meta.root_nodes, &mut f) |
|
527 | recur(on_disk, meta.root_nodes, &mut f) | |
526 | } |
|
528 | } | |
527 |
|
529 | |||
528 | /// Returns new data and metadata, together with whether that data should be |
|
530 | /// Returns new data and metadata, together with whether that data should be | |
529 | /// appended to the existing data file whose content is at |
|
531 | /// appended to the existing data file whose content is at | |
530 | /// `dirstate_map.on_disk` (true), instead of written to a new data file |
|
532 | /// `dirstate_map.on_disk` (true), instead of written to a new data file | |
531 | /// (false). |
|
533 | /// (false). | |
532 | pub(super) fn write( |
|
534 | pub(super) fn write( | |
533 | dirstate_map: &mut DirstateMap, |
|
535 | dirstate_map: &mut DirstateMap, | |
534 | can_append: bool, |
|
536 | can_append: bool, | |
535 | ) -> Result<(Vec<u8>, Vec<u8>, bool), DirstateError> { |
|
537 | ) -> Result<(Vec<u8>, Vec<u8>, bool), DirstateError> { | |
536 | let append = can_append && dirstate_map.write_should_append(); |
|
538 | let append = can_append && dirstate_map.write_should_append(); | |
537 |
|
539 | |||
538 | // This ignores the space for paths, and for nodes without an entry. |
|
540 | // This ignores the space for paths, and for nodes without an entry. | |
539 | // TODO: better estimate? Skip the `Vec` and write to a file directly? |
|
541 | // TODO: better estimate? Skip the `Vec` and write to a file directly? | |
540 | let size_guess = std::mem::size_of::<Node>() |
|
542 | let size_guess = std::mem::size_of::<Node>() | |
541 | * dirstate_map.nodes_with_entry_count as usize; |
|
543 | * dirstate_map.nodes_with_entry_count as usize; | |
542 |
|
544 | |||
543 | let mut writer = Writer { |
|
545 | let mut writer = Writer { | |
544 | dirstate_map, |
|
546 | dirstate_map, | |
545 | append, |
|
547 | append, | |
546 | out: Vec::with_capacity(size_guess), |
|
548 | out: Vec::with_capacity(size_guess), | |
547 | }; |
|
549 | }; | |
548 |
|
550 | |||
549 | let root_nodes = writer.write_nodes(dirstate_map.root.as_ref())?; |
|
551 | let root_nodes = writer.write_nodes(dirstate_map.root.as_ref())?; | |
550 |
|
552 | |||
551 | let meta = TreeMetadata { |
|
553 | let meta = TreeMetadata { | |
552 | root_nodes, |
|
554 | root_nodes, | |
553 | nodes_with_entry_count: dirstate_map.nodes_with_entry_count.into(), |
|
555 | nodes_with_entry_count: dirstate_map.nodes_with_entry_count.into(), | |
554 | nodes_with_copy_source_count: dirstate_map |
|
556 | nodes_with_copy_source_count: dirstate_map | |
555 | .nodes_with_copy_source_count |
|
557 | .nodes_with_copy_source_count | |
556 | .into(), |
|
558 | .into(), | |
557 | unreachable_bytes: dirstate_map.unreachable_bytes.into(), |
|
559 | unreachable_bytes: dirstate_map.unreachable_bytes.into(), | |
558 | unused: [0; 4], |
|
560 | unused: [0; 4], | |
559 | ignore_patterns_hash: dirstate_map.ignore_patterns_hash, |
|
561 | ignore_patterns_hash: dirstate_map.ignore_patterns_hash, | |
560 | }; |
|
562 | }; | |
561 | Ok((writer.out, meta.as_bytes().to_vec(), append)) |
|
563 | Ok((writer.out, meta.as_bytes().to_vec(), append)) | |
562 | } |
|
564 | } | |
563 |
|
565 | |||
564 | struct Writer<'dmap, 'on_disk> { |
|
566 | struct Writer<'dmap, 'on_disk> { | |
565 | dirstate_map: &'dmap DirstateMap<'on_disk>, |
|
567 | dirstate_map: &'dmap DirstateMap<'on_disk>, | |
566 | append: bool, |
|
568 | append: bool, | |
567 | out: Vec<u8>, |
|
569 | out: Vec<u8>, | |
568 | } |
|
570 | } | |
569 |
|
571 | |||
570 | impl Writer<'_, '_> { |
|
572 | impl Writer<'_, '_> { | |
571 | fn write_nodes( |
|
573 | fn write_nodes( | |
572 | &mut self, |
|
574 | &mut self, | |
573 | nodes: dirstate_map::ChildNodesRef, |
|
575 | nodes: dirstate_map::ChildNodesRef, | |
574 | ) -> Result<ChildNodes, DirstateError> { |
|
576 | ) -> Result<ChildNodes, DirstateError> { | |
575 | // Reuse already-written nodes if possible |
|
577 | // Reuse already-written nodes if possible | |
576 | if self.append { |
|
578 | if self.append { | |
577 | if let dirstate_map::ChildNodesRef::OnDisk(nodes_slice) = nodes { |
|
579 | if let dirstate_map::ChildNodesRef::OnDisk(nodes_slice) = nodes { | |
578 | let start = self.on_disk_offset_of(nodes_slice).expect( |
|
580 | let start = self.on_disk_offset_of(nodes_slice).expect( | |
579 | "dirstate-v2 OnDisk nodes not found within on_disk", |
|
581 | "dirstate-v2 OnDisk nodes not found within on_disk", | |
580 | ); |
|
582 | ); | |
581 | let len = child_nodes_len_from_usize(nodes_slice.len()); |
|
583 | let len = child_nodes_len_from_usize(nodes_slice.len()); | |
582 | return Ok(ChildNodes { start, len }); |
|
584 | return Ok(ChildNodes { start, len }); | |
583 | } |
|
585 | } | |
584 | } |
|
586 | } | |
585 |
|
587 | |||
586 | // `dirstate_map::ChildNodes::InMemory` contains a `HashMap` which has |
|
588 | // `dirstate_map::ChildNodes::InMemory` contains a `HashMap` which has | |
587 | // undefined iteration order. Sort to enable binary search in the |
|
589 | // undefined iteration order. Sort to enable binary search in the | |
588 | // written file. |
|
590 | // written file. | |
589 | let nodes = nodes.sorted(); |
|
591 | let nodes = nodes.sorted(); | |
590 | let nodes_len = nodes.len(); |
|
592 | let nodes_len = nodes.len(); | |
591 |
|
593 | |||
592 | // First accumulate serialized nodes in a `Vec` |
|
594 | // First accumulate serialized nodes in a `Vec` | |
593 | let mut on_disk_nodes = Vec::with_capacity(nodes_len); |
|
595 | let mut on_disk_nodes = Vec::with_capacity(nodes_len); | |
594 | for node in nodes { |
|
596 | for node in nodes { | |
595 | let children = |
|
597 | let children = | |
596 | self.write_nodes(node.children(self.dirstate_map.on_disk)?)?; |
|
598 | self.write_nodes(node.children(self.dirstate_map.on_disk)?)?; | |
597 | let full_path = node.full_path(self.dirstate_map.on_disk)?; |
|
599 | let full_path = node.full_path(self.dirstate_map.on_disk)?; | |
598 | let full_path = self.write_path(full_path.as_bytes()); |
|
600 | let full_path = self.write_path(full_path.as_bytes()); | |
599 | let copy_source = if let Some(source) = |
|
601 | let copy_source = if let Some(source) = | |
600 | node.copy_source(self.dirstate_map.on_disk)? |
|
602 | node.copy_source(self.dirstate_map.on_disk)? | |
601 | { |
|
603 | { | |
602 | self.write_path(source.as_bytes()) |
|
604 | self.write_path(source.as_bytes()) | |
603 | } else { |
|
605 | } else { | |
604 | PathSlice { |
|
606 | PathSlice { | |
605 | start: 0.into(), |
|
607 | start: 0.into(), | |
606 | len: 0.into(), |
|
608 | len: 0.into(), | |
607 | } |
|
609 | } | |
608 | }; |
|
610 | }; | |
609 | on_disk_nodes.push(match node { |
|
611 | on_disk_nodes.push(match node { | |
610 | NodeRef::InMemory(path, node) => { |
|
612 | NodeRef::InMemory(path, node) => { | |
611 | let (flags, data) = match &node.data { |
|
613 | let (flags, data) = match &node.data { | |
612 | dirstate_map::NodeData::Entry(entry) => { |
|
614 | dirstate_map::NodeData::Entry(entry) => { | |
613 | Entry::from_dirstate_entry(entry) |
|
615 | Entry::from_dirstate_entry(entry) | |
614 | } |
|
616 | } | |
615 | dirstate_map::NodeData::CachedDirectory { mtime } => { |
|
617 | dirstate_map::NodeData::CachedDirectory { mtime } => { | |
616 | (Flags::HAS_MTIME, Entry::from_timestamp(*mtime)) |
|
618 | (Flags::HAS_MTIME, Entry::from_timestamp(*mtime)) | |
617 | } |
|
619 | } | |
618 | dirstate_map::NodeData::None => ( |
|
620 | dirstate_map::NodeData::None => ( | |
619 | Flags::empty(), |
|
621 | Flags::empty(), | |
620 | Entry { |
|
622 | Entry { | |
621 | _padding: 0.into(), |
|
623 | _padding: 0.into(), | |
622 | size: 0.into(), |
|
624 | size: 0.into(), | |
623 | mtime: 0.into(), |
|
625 | mtime: 0.into(), | |
624 | }, |
|
626 | }, | |
625 | ), |
|
627 | ), | |
626 | }; |
|
628 | }; | |
627 | Node { |
|
629 | Node { | |
628 | children, |
|
630 | children, | |
629 | copy_source, |
|
631 | copy_source, | |
630 | full_path, |
|
632 | full_path, | |
631 | base_name_start: u16::try_from(path.base_name_start()) |
|
633 | base_name_start: u16::try_from(path.base_name_start()) | |
632 | // Could only panic for paths over 64 KiB |
|
634 | // Could only panic for paths over 64 KiB | |
633 | .expect("dirstate-v2 path length overflow") |
|
635 | .expect("dirstate-v2 path length overflow") | |
634 | .into(), |
|
636 | .into(), | |
635 | descendants_with_entry_count: node |
|
637 | descendants_with_entry_count: node | |
636 | .descendants_with_entry_count |
|
638 | .descendants_with_entry_count | |
637 | .into(), |
|
639 | .into(), | |
638 | tracked_descendants_count: node |
|
640 | tracked_descendants_count: node | |
639 | .tracked_descendants_count |
|
641 | .tracked_descendants_count | |
640 | .into(), |
|
642 | .into(), | |
641 | flags, |
|
643 | flags, | |
642 | data, |
|
644 | data, | |
643 | } |
|
645 | } | |
644 | } |
|
646 | } | |
645 | NodeRef::OnDisk(node) => Node { |
|
647 | NodeRef::OnDisk(node) => Node { | |
646 | children, |
|
648 | children, | |
647 | copy_source, |
|
649 | copy_source, | |
648 | full_path, |
|
650 | full_path, | |
649 | ..*node |
|
651 | ..*node | |
650 | }, |
|
652 | }, | |
651 | }) |
|
653 | }) | |
652 | } |
|
654 | } | |
653 | // β¦ so we can write them contiguously, after writing everything else |
|
655 | // β¦ so we can write them contiguously, after writing everything else | |
654 | // they refer to. |
|
656 | // they refer to. | |
655 | let start = self.current_offset(); |
|
657 | let start = self.current_offset(); | |
656 | let len = child_nodes_len_from_usize(nodes_len); |
|
658 | let len = child_nodes_len_from_usize(nodes_len); | |
657 | self.out.extend(on_disk_nodes.as_bytes()); |
|
659 | self.out.extend(on_disk_nodes.as_bytes()); | |
658 | Ok(ChildNodes { start, len }) |
|
660 | Ok(ChildNodes { start, len }) | |
659 | } |
|
661 | } | |
660 |
|
662 | |||
661 | /// If the given slice of items is within `on_disk`, returns its offset |
|
663 | /// If the given slice of items is within `on_disk`, returns its offset | |
662 | /// from the start of `on_disk`. |
|
664 | /// from the start of `on_disk`. | |
663 | fn on_disk_offset_of<T>(&self, slice: &[T]) -> Option<Offset> |
|
665 | fn on_disk_offset_of<T>(&self, slice: &[T]) -> Option<Offset> | |
664 | where |
|
666 | where | |
665 | T: BytesCast, |
|
667 | T: BytesCast, | |
666 | { |
|
668 | { | |
667 | fn address_range(slice: &[u8]) -> std::ops::RangeInclusive<usize> { |
|
669 | fn address_range(slice: &[u8]) -> std::ops::RangeInclusive<usize> { | |
668 | let start = slice.as_ptr() as usize; |
|
670 | let start = slice.as_ptr() as usize; | |
669 | let end = start + slice.len(); |
|
671 | let end = start + slice.len(); | |
670 | start..=end |
|
672 | start..=end | |
671 | } |
|
673 | } | |
672 | let slice_addresses = address_range(slice.as_bytes()); |
|
674 | let slice_addresses = address_range(slice.as_bytes()); | |
673 | let on_disk_addresses = address_range(self.dirstate_map.on_disk); |
|
675 | let on_disk_addresses = address_range(self.dirstate_map.on_disk); | |
674 | if on_disk_addresses.contains(slice_addresses.start()) |
|
676 | if on_disk_addresses.contains(slice_addresses.start()) | |
675 | && on_disk_addresses.contains(slice_addresses.end()) |
|
677 | && on_disk_addresses.contains(slice_addresses.end()) | |
676 | { |
|
678 | { | |
677 | let offset = slice_addresses.start() - on_disk_addresses.start(); |
|
679 | let offset = slice_addresses.start() - on_disk_addresses.start(); | |
678 | Some(offset_from_usize(offset)) |
|
680 | Some(offset_from_usize(offset)) | |
679 | } else { |
|
681 | } else { | |
680 | None |
|
682 | None | |
681 | } |
|
683 | } | |
682 | } |
|
684 | } | |
683 |
|
685 | |||
684 | fn current_offset(&mut self) -> Offset { |
|
686 | fn current_offset(&mut self) -> Offset { | |
685 | let mut offset = self.out.len(); |
|
687 | let mut offset = self.out.len(); | |
686 | if self.append { |
|
688 | if self.append { | |
687 | offset += self.dirstate_map.on_disk.len() |
|
689 | offset += self.dirstate_map.on_disk.len() | |
688 | } |
|
690 | } | |
689 | offset_from_usize(offset) |
|
691 | offset_from_usize(offset) | |
690 | } |
|
692 | } | |
691 |
|
693 | |||
692 | fn write_path(&mut self, slice: &[u8]) -> PathSlice { |
|
694 | fn write_path(&mut self, slice: &[u8]) -> PathSlice { | |
693 | let len = path_len_from_usize(slice.len()); |
|
695 | let len = path_len_from_usize(slice.len()); | |
694 | // Reuse an already-written path if possible |
|
696 | // Reuse an already-written path if possible | |
695 | if self.append { |
|
697 | if self.append { | |
696 | if let Some(start) = self.on_disk_offset_of(slice) { |
|
698 | if let Some(start) = self.on_disk_offset_of(slice) { | |
697 | return PathSlice { start, len }; |
|
699 | return PathSlice { start, len }; | |
698 | } |
|
700 | } | |
699 | } |
|
701 | } | |
700 | let start = self.current_offset(); |
|
702 | let start = self.current_offset(); | |
701 | self.out.extend(slice.as_bytes()); |
|
703 | self.out.extend(slice.as_bytes()); | |
702 | PathSlice { start, len } |
|
704 | PathSlice { start, len } | |
703 | } |
|
705 | } | |
704 | } |
|
706 | } | |
705 |
|
707 | |||
706 | fn offset_from_usize(x: usize) -> Offset { |
|
708 | fn offset_from_usize(x: usize) -> Offset { | |
707 | u32::try_from(x) |
|
709 | u32::try_from(x) | |
708 | // Could only panic for a dirstate file larger than 4 GiB |
|
710 | // Could only panic for a dirstate file larger than 4 GiB | |
709 | .expect("dirstate-v2 offset overflow") |
|
711 | .expect("dirstate-v2 offset overflow") | |
710 | .into() |
|
712 | .into() | |
711 | } |
|
713 | } | |
712 |
|
714 | |||
713 | fn child_nodes_len_from_usize(x: usize) -> Size { |
|
715 | fn child_nodes_len_from_usize(x: usize) -> Size { | |
714 | u32::try_from(x) |
|
716 | u32::try_from(x) | |
715 | // Could only panic with over 4 billion nodes |
|
717 | // Could only panic with over 4 billion nodes | |
716 | .expect("dirstate-v2 slice length overflow") |
|
718 | .expect("dirstate-v2 slice length overflow") | |
717 | .into() |
|
719 | .into() | |
718 | } |
|
720 | } | |
719 |
|
721 | |||
720 | fn path_len_from_usize(x: usize) -> PathSize { |
|
722 | fn path_len_from_usize(x: usize) -> PathSize { | |
721 | u16::try_from(x) |
|
723 | u16::try_from(x) | |
722 | // Could only panic for paths over 64 KiB |
|
724 | // Could only panic for paths over 64 KiB | |
723 | .expect("dirstate-v2 path length overflow") |
|
725 | .expect("dirstate-v2 path length overflow") | |
724 | .into() |
|
726 | .into() | |
725 | } |
|
727 | } |
General Comments 0
You need to be logged in to leave comments.
Login now