| ... | ... | @@ -44,20 +44,23 @@ pub const Symbol = struct { |
| 44 | 44 | export_flags: u64, |
| 45 | 45 | }; |
| 46 | 46 | |
| 47 | | const Edge = struct { |
| 47 | pub const Edge = struct { |
| 48 | 48 | from: *Node, |
| 49 | 49 | to: *Node, |
| 50 | | label: []const u8, |
| 50 | label: []u8, |
| 51 | 51 | |
| 52 | | fn deinit(self: *Edge, alloc: *Allocator) void { |
| 53 | | self.to.deinit(alloc); |
| 54 | | alloc.destroy(self.to); |
| 52 | fn deinit(self: *Edge, allocator: *Allocator) void { |
| 53 | self.to.deinit(); |
| 54 | allocator.destroy(self.to); |
| 55 | allocator.free(self.label); |
| 55 | 56 | self.from = undefined; |
| 56 | 57 | self.to = undefined; |
| 58 | self.label = undefined; |
| 57 | 59 | } |
| 58 | 60 | }; |
| 59 | 61 | |
| 60 | | const Node = struct { |
| 62 | pub const Node = struct { |
| 63 | base: *Trie, |
| 61 | 64 | /// Export flags associated with this exported symbol (if any). |
| 62 | 65 | export_flags: ?u64 = null, |
| 63 | 66 | /// VM address offset wrt to the section this symbol is defined against (if any). |
| ... | ... | @@ -67,73 +70,97 @@ const Node = struct { |
| 67 | 70 | /// List of all edges originating from this node. |
| 68 | 71 | edges: std.ArrayListUnmanaged(Edge) = .{}, |
| 69 | 72 | |
| 70 | | fn deinit(self: *Node, alloc: *Allocator) void { |
| 73 | fn deinit(self: *Node) void { |
| 71 | 74 | for (self.edges.items) |*edge| { |
| 72 | | edge.deinit(alloc); |
| 75 | edge.deinit(self.base.allocator); |
| 73 | 76 | } |
| 74 | | self.edges.deinit(alloc); |
| 77 | self.edges.deinit(self.base.allocator); |
| 75 | 78 | } |
| 76 | 79 | |
| 77 | | const PutResult = struct { |
| 78 | | /// Node reached at this stage of `put` op. |
| 79 | | node: *Node, |
| 80 | | /// Count of newly inserted nodes at this stage of `put` op. |
| 81 | | node_count: usize, |
| 82 | | }; |
| 83 | | |
| 84 | 80 | /// Inserts a new node starting from `self`. |
| 85 | | fn put(self: *Node, alloc: *Allocator, label: []const u8, node_count: usize) !PutResult { |
| 86 | | var curr_node_count = node_count; |
| 81 | fn put(self: *Node, label: []const u8) !*Node { |
| 87 | 82 | // Check for match with edges from this node. |
| 88 | 83 | for (self.edges.items) |*edge| { |
| 89 | | const match = mem.indexOfDiff(u8, edge.label, label) orelse return PutResult{ |
| 90 | | .node = edge.to, |
| 91 | | .node_count = curr_node_count, |
| 92 | | }; |
| 84 | const match = mem.indexOfDiff(u8, edge.label, label) orelse return edge.to; |
| 93 | 85 | if (match == 0) continue; |
| 94 | | if (match == edge.label.len) return edge.to.put(alloc, label[match..], curr_node_count); |
| 86 | if (match == edge.label.len) return edge.to.put(label[match..]); |
| 95 | 87 | |
| 96 | 88 | // Found a match, need to splice up nodes. |
| 97 | 89 | // From: A -> B |
| 98 | 90 | // To: A -> C -> B |
| 99 | | const mid = try alloc.create(Node); |
| 100 | | mid.* = .{}; |
| 101 | | const to_label = edge.label; |
| 91 | const mid = try self.base.allocator.create(Node); |
| 92 | mid.* = .{ .base = self.base }; |
| 93 | var to_label = try self.base.allocator.dupe(u8, edge.label[match..]); |
| 94 | self.base.allocator.free(edge.label); |
| 102 | 95 | const to_node = edge.to; |
| 103 | 96 | edge.to = mid; |
| 104 | | edge.label = label[0..match]; |
| 105 | | curr_node_count += 1; |
| 97 | edge.label = try self.base.allocator.dupe(u8, label[0..match]); |
| 98 | self.base.node_count += 1; |
| 106 | 99 | |
| 107 | | try mid.edges.append(alloc, .{ |
| 100 | try mid.edges.append(self.base.allocator, .{ |
| 108 | 101 | .from = mid, |
| 109 | 102 | .to = to_node, |
| 110 | | .label = to_label[match..], |
| 103 | .label = to_label, |
| 111 | 104 | }); |
| 112 | 105 | |
| 113 | | if (match == label.len) { |
| 114 | | return PutResult{ .node = to_node, .node_count = curr_node_count }; |
| 115 | | } else { |
| 116 | | return mid.put(alloc, label[match..], curr_node_count); |
| 117 | | } |
| 106 | return if (match == label.len) to_node else mid.put(label[match..]); |
| 118 | 107 | } |
| 119 | 108 | |
| 120 | 109 | // Add a new node. |
| 121 | | const node = try alloc.create(Node); |
| 122 | | node.* = .{}; |
| 123 | | curr_node_count += 1; |
| 110 | const node = try self.base.allocator.create(Node); |
| 111 | node.* = .{ .base = self.base }; |
| 112 | self.base.node_count += 1; |
| 124 | 113 | |
| 125 | | try self.edges.append(alloc, .{ |
| 114 | try self.edges.append(self.base.allocator, .{ |
| 126 | 115 | .from = self, |
| 127 | 116 | .to = node, |
| 128 | | .label = label, |
| 117 | .label = try self.base.allocator.dupe(u8, label), |
| 129 | 118 | }); |
| 130 | 119 | |
| 131 | | return PutResult{ .node = node, .node_count = curr_node_count }; |
| 120 | return node; |
| 121 | } |
| 122 | |
| 123 | fn fromByteStream(self: *Node, stream: anytype) Trie.FromByteStreamError!void { |
| 124 | self.trie_offset = try stream.getPos(); |
| 125 | var reader = stream.reader(); |
| 126 | const node_size = try leb.readULEB128(u64, reader); |
| 127 | if (node_size > 0) { |
| 128 | self.export_flags = try leb.readULEB128(u64, reader); |
| 129 | // TODO Parse flags. |
| 130 | self.vmaddr_offset = try leb.readULEB128(u64, reader); |
| 131 | } |
| 132 | const nedges = try reader.readByte(); |
| 133 | self.base.node_count += nedges; |
| 134 | var i: usize = 0; |
| 135 | while (i < nedges) : (i += 1) { |
| 136 | var label = blk: { |
| 137 | var label_buf = std.ArrayList(u8).init(self.base.allocator); |
| 138 | while (true) { |
| 139 | const next = try reader.readByte(); |
| 140 | if (next == @as(u8, 0)) |
| 141 | break; |
| 142 | try label_buf.append(next); |
| 143 | } |
| 144 | break :blk label_buf.toOwnedSlice(); |
| 145 | }; |
| 146 | const seek_to = try leb.readULEB128(u64, reader); |
| 147 | const cur_pos = try stream.getPos(); |
| 148 | try stream.seekTo(seek_to); |
| 149 | var node = try self.base.allocator.create(Node); |
| 150 | node.* = .{ .base = self.base }; |
| 151 | try node.fromByteStream(stream); |
| 152 | try self.edges.append(self.base.allocator, .{ |
| 153 | .from = self, |
| 154 | .to = node, |
| 155 | .label = label, |
| 156 | }); |
| 157 | try stream.seekTo(cur_pos); |
| 158 | } |
| 132 | 159 | } |
| 133 | 160 | |
| 134 | 161 | /// This method should only be called *after* updateOffset has been called! |
| 135 | 162 | /// In case this is not upheld, this method will panic. |
| 136 | | fn writeULEB128Mem(self: Node, buffer: *std.ArrayListUnmanaged(u8)) !void { |
| 163 | fn writeULEB128Mem(self: Node, buffer: *std.ArrayList(u8)) !void { |
| 137 | 164 | assert(self.trie_offset != null); // You need to call updateOffset first. |
| 138 | 165 | if (self.vmaddr_offset) |offset| { |
| 139 | 166 | // Terminal node info: encode export flags and vmaddr offset of this symbol. |
| ... | ... | @@ -221,64 +248,95 @@ const Node = struct { |
| 221 | 248 | /// the count always starts at 1. |
| 222 | 249 | node_count: usize = 1, |
| 223 | 250 | /// The root node of the trie. |
| 224 | | root: Node = .{}, |
| 251 | root: ?Node = null, |
| 252 | allocator: *Allocator, |
| 253 | |
| 254 | pub fn init(allocator: *Allocator) Trie { |
| 255 | return .{ .allocator = allocator }; |
| 256 | } |
| 225 | 257 | |
| 226 | 258 | /// Insert a symbol into the trie, updating the prefixes in the process. |
| 227 | 259 | /// This operation may change the layout of the trie by splicing edges in |
| 228 | 260 | /// certain circumstances. |
| 229 | | pub fn put(self: *Trie, alloc: *Allocator, symbol: Symbol) !void { |
| 230 | | const res = try self.root.put(alloc, symbol.name, 0); |
| 231 | | self.node_count += res.node_count; |
| 232 | | res.node.vmaddr_offset = symbol.vmaddr_offset; |
| 233 | | res.node.export_flags = symbol.export_flags; |
| 261 | pub fn put(self: *Trie, symbol: Symbol) !void { |
| 262 | if (self.root == null) { |
| 263 | self.root = .{ .base = self }; |
| 264 | } |
| 265 | const node = try self.root.?.put(symbol.name); |
| 266 | node.vmaddr_offset = symbol.vmaddr_offset; |
| 267 | node.export_flags = symbol.export_flags; |
| 234 | 268 | } |
| 235 | 269 | |
| 236 | | /// Write the trie to a buffer ULEB128 encoded. |
| 237 | | pub fn writeULEB128Mem(self: *Trie, alloc: *Allocator, buffer: *std.ArrayListUnmanaged(u8)) !void { |
| 238 | | var ordered_nodes: std.ArrayListUnmanaged(*Node) = .{}; |
| 239 | | defer ordered_nodes.deinit(alloc); |
| 270 | const FromByteStreamError = error{ |
| 271 | OutOfMemory, |
| 272 | EndOfStream, |
| 273 | Overflow, |
| 274 | }; |
| 240 | 275 | |
| 241 | | try ordered_nodes.ensureCapacity(alloc, self.node_count); |
| 242 | | walkInOrder(&self.root, &ordered_nodes); |
| 276 | /// Parse the trie from a byte stream. |
| 277 | pub fn fromByteStream(self: *Trie, stream: anytype) FromByteStreamError!void { |
| 278 | if (self.root == null) { |
| 279 | self.root = .{ .base = self }; |
| 280 | } |
| 281 | return self.root.?.fromByteStream(stream); |
| 282 | } |
| 283 | |
| 284 | /// Write the trie to a buffer ULEB128 encoded. |
| 285 | /// Caller owns the memory and needs to free it. |
| 286 | pub fn writeULEB128Mem(self: *Trie) ![]u8 { |
| 287 | var ordered_nodes = try self.nodes(); |
| 288 | defer self.allocator.free(ordered_nodes); |
| 243 | 289 | |
| 244 | 290 | var offset: usize = 0; |
| 245 | 291 | var more: bool = true; |
| 246 | 292 | while (more) { |
| 247 | 293 | offset = 0; |
| 248 | 294 | more = false; |
| 249 | | for (ordered_nodes.items) |node| { |
| 295 | for (ordered_nodes) |node| { |
| 250 | 296 | const res = node.updateOffset(offset); |
| 251 | 297 | offset += res.node_size; |
| 252 | 298 | if (res.updated) more = true; |
| 253 | 299 | } |
| 254 | 300 | } |
| 255 | 301 | |
| 256 | | try buffer.ensureCapacity(alloc, buffer.items.len + offset); |
| 257 | | for (ordered_nodes.items) |node| { |
| 258 | | try node.writeULEB128Mem(buffer); |
| 302 | var buffer = std.ArrayList(u8).init(self.allocator); |
| 303 | try buffer.ensureCapacity(offset); |
| 304 | for (ordered_nodes) |node| { |
| 305 | try node.writeULEB128Mem(&buffer); |
| 259 | 306 | } |
| 307 | return buffer.toOwnedSlice(); |
| 260 | 308 | } |
| 261 | 309 | |
| 262 | | /// Walks the trie in DFS order gathering all nodes into a linear stream of nodes. |
| 263 | | fn walkInOrder(node: *Node, list: *std.ArrayListUnmanaged(*Node)) void { |
| 264 | | list.appendAssumeCapacity(node); |
| 265 | | for (node.edges.items) |*edge| { |
| 266 | | walkInOrder(edge.to, list); |
| 310 | pub fn nodes(self: *Trie) ![]*Node { |
| 311 | var ordered_nodes = std.ArrayList(*Node).init(self.allocator); |
| 312 | try ordered_nodes.ensureCapacity(self.node_count); |
| 313 | |
| 314 | comptime const Fifo = std.fifo.LinearFifo(*Node, .{ .Static = std.math.maxInt(u8) }); |
| 315 | var fifo = Fifo.init(); |
| 316 | try fifo.writeItem(&self.root.?); |
| 317 | |
| 318 | while (fifo.readItem()) |next| { |
| 319 | for (next.edges.items) |*edge| { |
| 320 | try fifo.writeItem(edge.to); |
| 321 | } |
| 322 | ordered_nodes.appendAssumeCapacity(next); |
| 267 | 323 | } |
| 324 | |
| 325 | return ordered_nodes.toOwnedSlice(); |
| 268 | 326 | } |
| 269 | 327 | |
| 270 | | pub fn deinit(self: *Trie, alloc: *Allocator) void { |
| 271 | | self.root.deinit(alloc); |
| 328 | pub fn deinit(self: *Trie) void { |
| 329 | self.root.?.deinit(); |
| 272 | 330 | } |
| 273 | 331 | |
| 274 | 332 | test "Trie node count" { |
| 275 | 333 | var gpa = testing.allocator; |
| 276 | | var trie: Trie = .{}; |
| 277 | | defer trie.deinit(gpa); |
| 334 | var trie = Trie.init(gpa); |
| 335 | defer trie.deinit(); |
| 278 | 336 | |
| 279 | 337 | testing.expectEqual(trie.node_count, 1); |
| 280 | 338 | |
| 281 | | try trie.put(gpa, .{ |
| 339 | try trie.put(.{ |
| 282 | 340 | .name = "_main", |
| 283 | 341 | .vmaddr_offset = 0, |
| 284 | 342 | .export_flags = 0, |
| ... | ... | @@ -286,14 +344,14 @@ test "Trie node count" { |
| 286 | 344 | testing.expectEqual(trie.node_count, 2); |
| 287 | 345 | |
| 288 | 346 | // Inserting the same node shouldn't update the trie. |
| 289 | | try trie.put(gpa, .{ |
| 347 | try trie.put(.{ |
| 290 | 348 | .name = "_main", |
| 291 | 349 | .vmaddr_offset = 0, |
| 292 | 350 | .export_flags = 0, |
| 293 | 351 | }); |
| 294 | 352 | testing.expectEqual(trie.node_count, 2); |
| 295 | 353 | |
| 296 | | try trie.put(gpa, .{ |
| 354 | try trie.put(.{ |
| 297 | 355 | .name = "__mh_execute_header", |
| 298 | 356 | .vmaddr_offset = 0x1000, |
| 299 | 357 | .export_flags = 0, |
| ... | ... | @@ -301,13 +359,13 @@ test "Trie node count" { |
| 301 | 359 | testing.expectEqual(trie.node_count, 4); |
| 302 | 360 | |
| 303 | 361 | // Inserting the same node shouldn't update the trie. |
| 304 | | try trie.put(gpa, .{ |
| 362 | try trie.put(.{ |
| 305 | 363 | .name = "__mh_execute_header", |
| 306 | 364 | .vmaddr_offset = 0x1000, |
| 307 | 365 | .export_flags = 0, |
| 308 | 366 | }); |
| 309 | 367 | testing.expectEqual(trie.node_count, 4); |
| 310 | | try trie.put(gpa, .{ |
| 368 | try trie.put(.{ |
| 311 | 369 | .name = "_main", |
| 312 | 370 | .vmaddr_offset = 0, |
| 313 | 371 | .export_flags = 0, |
| ... | ... | @@ -317,31 +375,28 @@ test "Trie node count" { |
| 317 | 375 | |
| 318 | 376 | test "Trie basic" { |
| 319 | 377 | var gpa = testing.allocator; |
| 320 | | var trie: Trie = .{}; |
| 321 | | defer trie.deinit(gpa); |
| 322 | | |
| 323 | | // root |
| 324 | | testing.expect(trie.root.edges.items.len == 0); |
| 378 | var trie = Trie.init(gpa); |
| 379 | defer trie.deinit(); |
| 325 | 380 | |
| 326 | 381 | // root --- _st ---> node |
| 327 | | try trie.put(gpa, .{ |
| 382 | try trie.put(.{ |
| 328 | 383 | .name = "_st", |
| 329 | 384 | .vmaddr_offset = 0, |
| 330 | 385 | .export_flags = 0, |
| 331 | 386 | }); |
| 332 | | testing.expect(trie.root.edges.items.len == 1); |
| 333 | | testing.expect(mem.eql(u8, trie.root.edges.items[0].label, "_st")); |
| 387 | testing.expect(trie.root.?.edges.items.len == 1); |
| 388 | testing.expect(mem.eql(u8, trie.root.?.edges.items[0].label, "_st")); |
| 334 | 389 | |
| 335 | 390 | { |
| 336 | 391 | // root --- _st ---> node --- art ---> node |
| 337 | | try trie.put(gpa, .{ |
| 392 | try trie.put(.{ |
| 338 | 393 | .name = "_start", |
| 339 | 394 | .vmaddr_offset = 0, |
| 340 | 395 | .export_flags = 0, |
| 341 | 396 | }); |
| 342 | | testing.expect(trie.root.edges.items.len == 1); |
| 397 | testing.expect(trie.root.?.edges.items.len == 1); |
| 343 | 398 | |
| 344 | | const nextEdge = &trie.root.edges.items[0]; |
| 399 | const nextEdge = &trie.root.?.edges.items[0]; |
| 345 | 400 | testing.expect(mem.eql(u8, nextEdge.label, "_st")); |
| 346 | 401 | testing.expect(nextEdge.to.edges.items.len == 1); |
| 347 | 402 | testing.expect(mem.eql(u8, nextEdge.to.edges.items[0].label, "art")); |
| ... | ... | @@ -350,14 +405,14 @@ test "Trie basic" { |
| 350 | 405 | // root --- _ ---> node --- st ---> node --- art ---> node |
| 351 | 406 | // | |
| 352 | 407 | // | --- main ---> node |
| 353 | | try trie.put(gpa, .{ |
| 408 | try trie.put(.{ |
| 354 | 409 | .name = "_main", |
| 355 | 410 | .vmaddr_offset = 0, |
| 356 | 411 | .export_flags = 0, |
| 357 | 412 | }); |
| 358 | | testing.expect(trie.root.edges.items.len == 1); |
| 413 | testing.expect(trie.root.?.edges.items.len == 1); |
| 359 | 414 | |
| 360 | | const nextEdge = &trie.root.edges.items[0]; |
| 415 | const nextEdge = &trie.root.?.edges.items[0]; |
| 361 | 416 | testing.expect(mem.eql(u8, nextEdge.label, "_")); |
| 362 | 417 | testing.expect(nextEdge.to.edges.items.len == 2); |
| 363 | 418 | testing.expect(mem.eql(u8, nextEdge.to.edges.items[0].label, "st")); |
| ... | ... | @@ -370,24 +425,22 @@ test "Trie basic" { |
| 370 | 425 | |
| 371 | 426 | test "Trie.writeULEB128Mem" { |
| 372 | 427 | var gpa = testing.allocator; |
| 373 | | var trie: Trie = .{}; |
| 374 | | defer trie.deinit(gpa); |
| 428 | var trie = Trie.init(gpa); |
| 429 | defer trie.deinit(); |
| 375 | 430 | |
| 376 | | try trie.put(gpa, .{ |
| 431 | try trie.put(.{ |
| 377 | 432 | .name = "__mh_execute_header", |
| 378 | 433 | .vmaddr_offset = 0, |
| 379 | 434 | .export_flags = 0, |
| 380 | 435 | }); |
| 381 | | try trie.put(gpa, .{ |
| 436 | try trie.put(.{ |
| 382 | 437 | .name = "_main", |
| 383 | 438 | .vmaddr_offset = 0x1000, |
| 384 | 439 | .export_flags = 0, |
| 385 | 440 | }); |
| 386 | 441 | |
| 387 | | var buffer: std.ArrayListUnmanaged(u8) = .{}; |
| 388 | | defer buffer.deinit(gpa); |
| 389 | | |
| 390 | | try trie.writeULEB128Mem(gpa, &buffer); |
| 442 | var buffer = try trie.writeULEB128Mem(); |
| 443 | defer gpa.free(buffer); |
| 391 | 444 | |
| 392 | 445 | const exp_buffer = [_]u8{ |
| 393 | 446 | 0x0, |
| ... | ... | @@ -434,6 +487,64 @@ test "Trie.writeULEB128Mem" { |
| 434 | 487 | 0x0, |
| 435 | 488 | }; |
| 436 | 489 | |
| 437 | | testing.expect(buffer.items.len == exp_buffer.len); |
| 438 | | testing.expect(mem.eql(u8, buffer.items, exp_buffer[0..])); |
| 490 | testing.expect(buffer.len == exp_buffer.len); |
| 491 | testing.expect(mem.eql(u8, buffer, exp_buffer[0..])); |
| 492 | } |
| 493 | |
| 494 | test "parse Trie from byte stream" { |
| 495 | var gpa = testing.allocator; |
| 496 | |
| 497 | const in_buffer = [_]u8{ |
| 498 | 0x0, |
| 499 | 0x1, |
| 500 | 0x5f, |
| 501 | 0x0, |
| 502 | 0x5, |
| 503 | 0x0, |
| 504 | 0x2, |
| 505 | 0x5f, |
| 506 | 0x6d, |
| 507 | 0x68, |
| 508 | 0x5f, |
| 509 | 0x65, |
| 510 | 0x78, |
| 511 | 0x65, |
| 512 | 0x63, |
| 513 | 0x75, |
| 514 | 0x74, |
| 515 | 0x65, |
| 516 | 0x5f, |
| 517 | 0x68, |
| 518 | 0x65, |
| 519 | 0x61, |
| 520 | 0x64, |
| 521 | 0x65, |
| 522 | 0x72, |
| 523 | 0x0, |
| 524 | 0x21, |
| 525 | 0x6d, |
| 526 | 0x61, |
| 527 | 0x69, |
| 528 | 0x6e, |
| 529 | 0x0, |
| 530 | 0x25, |
| 531 | 0x2, |
| 532 | 0x0, |
| 533 | 0x0, |
| 534 | 0x0, |
| 535 | 0x3, |
| 536 | 0x0, |
| 537 | 0x80, |
| 538 | 0x20, |
| 539 | 0x0, |
| 540 | }; |
| 541 | var stream = std.io.fixedBufferStream(in_buffer[0..]); |
| 542 | var trie = Trie.init(gpa); |
| 543 | defer trie.deinit(); |
| 544 | try trie.fromByteStream(&stream); |
| 545 | |
| 546 | var out_buffer = try trie.writeULEB128Mem(); |
| 547 | defer gpa.free(out_buffer); |
| 548 | |
| 549 | testing.expect(mem.eql(u8, in_buffer[0..], out_buffer)); |
| 439 | 550 | } |