| ... | @@ -17,7 +17,8 @@ fn canDeduplicate(opcode: Opcode) bool { | ... | @@ -17,7 +17,8 @@ fn canDeduplicate(opcode: Opcode) bool { |
| 17 | // These are deprecated, so don't bother supporting them for now. | 17 | // These are deprecated, so don't bother supporting them for now. |
| 18 | return false; | 18 | return false; |
| 19 | }, | 19 | }, |
| 20 | .OpName, .OpMemberName => true, // Debug decoration-style instructions | 20 | // Debug decoration-style instructions |
| | 21 | .OpName, .OpMemberName => true, |
| 21 | else => switch (opcode.class()) { | 22 | else => switch (opcode.class()) { |
| 22 | .TypeDeclaration, | 23 | .TypeDeclaration, |
| 23 | .ConstantCreation, | 24 | .ConstantCreation, |
| ... | @@ -44,6 +45,8 @@ const ModuleInfo = struct { | ... | @@ -44,6 +45,8 @@ const ModuleInfo = struct { |
| 44 | /// or the entity that is affected by this entity if this entity | 45 | /// or the entity that is affected by this entity if this entity |
| 45 | /// is a decoration. | 46 | /// is a decoration. |
| 46 | result_id_index: u16, | 47 | result_id_index: u16, |
| | 48 | /// The first decoration in `self.decorations`. |
| | 49 | first_decoration: u32, |
| 47 | }; | 50 | }; |
| 48 | | 51 | |
| 49 | /// Maps result-id to Entity's | 52 | /// Maps result-id to Entity's |
| ... | @@ -53,6 +56,8 @@ const ModuleInfo = struct { | ... | @@ -53,6 +56,8 @@ const ModuleInfo = struct { |
| 53 | /// Because we need these values when recoding the module anyway, | 56 | /// Because we need these values when recoding the module anyway, |
| 54 | /// it contains the status of ALL operands in the module. | 57 | /// it contains the status of ALL operands in the module. |
| 55 | operand_is_id: std.DynamicBitSetUnmanaged, | 58 | operand_is_id: std.DynamicBitSetUnmanaged, |
| | 59 | /// Store of decorations for each entity. |
| | 60 | decorations: []const Entity, |
| 56 | | 61 | |
| 57 | pub fn parse( | 62 | pub fn parse( |
| 58 | arena: Allocator, | 63 | arena: Allocator, |
| ... | @@ -62,6 +67,7 @@ const ModuleInfo = struct { | ... | @@ -62,6 +67,7 @@ const ModuleInfo = struct { |
| 62 | var entities = std.AutoArrayHashMap(ResultId, Entity).init(arena); | 67 | var entities = std.AutoArrayHashMap(ResultId, Entity).init(arena); |
| 63 | var id_offsets = std.ArrayList(u16).init(arena); | 68 | var id_offsets = std.ArrayList(u16).init(arena); |
| 64 | var operand_is_id = try std.DynamicBitSetUnmanaged.initEmpty(arena, binary.instructions.len); | 69 | var operand_is_id = try std.DynamicBitSetUnmanaged.initEmpty(arena, binary.instructions.len); |
| | 70 | var decorations = std.MultiArrayList(struct { target_id: ResultId, entity: Entity }){}; |
| 65 | | 71 | |
| 66 | var it = binary.iterateInstructions(); | 72 | var it = binary.iterateInstructions(); |
| 67 | while (it.next()) |inst| { | 73 | while (it.next()) |inst| { |
| ... | @@ -82,10 +88,20 @@ const ModuleInfo = struct { | ... | @@ -82,10 +88,20 @@ const ModuleInfo = struct { |
| 82 | }; | 88 | }; |
| 83 | | 89 | |
| 84 | const result_id: ResultId = @enumFromInt(inst.operands[id_offsets.items[result_id_index]]); | 90 | const result_id: ResultId = @enumFromInt(inst.operands[id_offsets.items[result_id_index]]); |
| | 91 | const entity = Entity{ |
| | 92 | .kind = inst.opcode, |
| | 93 | .first_operand = first_operand_offset, |
| | 94 | .num_operands = @intCast(inst.operands.len), |
| | 95 | .result_id_index = result_id_index, |
| | 96 | .first_decoration = undefined, // Filled in later |
| | 97 | }; |
| 85 | | 98 | |
| 86 | switch (inst.opcode.class()) { | 99 | switch (inst.opcode.class()) { |
| 87 | .Annotation, .Debug => { | 100 | .Annotation, .Debug => { |
| 88 | // TODO | 101 | try decorations.append(arena, .{ |
| | 102 | .target_id = result_id, |
| | 103 | .entity = entity, |
| | 104 | }); |
| 89 | }, | 105 | }, |
| 90 | .TypeDeclaration, .ConstantCreation => { | 106 | .TypeDeclaration, .ConstantCreation => { |
| 91 | const entry = try entities.getOrPut(result_id); | 107 | const entry = try entities.getOrPut(result_id); |
| ... | @@ -93,22 +109,67 @@ const ModuleInfo = struct { | ... | @@ -93,22 +109,67 @@ const ModuleInfo = struct { |
| 93 | log.err("type or constant {} has duplicate definition", .{result_id}); | 109 | log.err("type or constant {} has duplicate definition", .{result_id}); |
| 94 | return error.DuplicateId; | 110 | return error.DuplicateId; |
| 95 | } | 111 | } |
| 96 | entry.value_ptr.* = .{ | 112 | entry.value_ptr.* = entity; |
| 97 | .kind = inst.opcode, | | |
| 98 | .first_operand = first_operand_offset, | | |
| 99 | .num_operands = @intCast(inst.operands.len), | | |
| 100 | .result_id_index = result_id_index, | | |
| 101 | }; | | |
| 102 | }, | 113 | }, |
| 103 | else => unreachable, | 114 | else => unreachable, |
| 104 | } | 115 | } |
| 105 | } | 116 | } |
| 106 | | 117 | |
| | 118 | // Sort decorations by the index of the result-id in `entities. |
| | 119 | // This ensures not only that the decorations of a particular reuslt-id |
| | 120 | // are continuous, but the subsequences also appear in the same order as in `entities`. |
| | 121 | |
| | 122 | const SortContext = struct { |
| | 123 | entities: std.AutoArrayHashMapUnmanaged(ResultId, Entity), |
| | 124 | ids: []const ResultId, |
| | 125 | |
| | 126 | pub fn lessThan(ctx: @This(), a_index: usize, b_index: usize) bool { |
| | 127 | // If any index is not in the entities set, its because its not a |
| | 128 | // deduplicatable result-id. Those should be considered largest and |
| | 129 | // float to the end. |
| | 130 | const entity_index_a = ctx.entities.getIndex(ctx.ids[a_index]) orelse return false; |
| | 131 | const entity_index_b = ctx.entities.getIndex(ctx.ids[b_index]) orelse return true; |
| | 132 | |
| | 133 | return entity_index_a < entity_index_b; |
| | 134 | } |
| | 135 | }; |
| | 136 | |
| | 137 | decorations.sort(SortContext{ |
| | 138 | .entities = entities.unmanaged, |
| | 139 | .ids = decorations.items(.target_id), |
| | 140 | }); |
| | 141 | |
| | 142 | // Now go through the decorations and add the offsets to the entities list. |
| | 143 | var decoration_i: u32 = 0; |
| | 144 | const target_ids = decorations.items(.target_id); |
| | 145 | for (entities.keys(), entities.values()) |id, *entity| { |
| | 146 | entity.first_decoration = decoration_i; |
| | 147 | |
| | 148 | // Scan ahead to the next decoration |
| | 149 | while (decoration_i < target_ids.len and target_ids[decoration_i] == id) { |
| | 150 | decoration_i += 1; |
| | 151 | } |
| | 152 | } |
| | 153 | |
| 107 | return ModuleInfo{ | 154 | return ModuleInfo{ |
| 108 | .entities = entities.unmanaged, | 155 | .entities = entities.unmanaged, |
| 109 | .operand_is_id = operand_is_id, | 156 | .operand_is_id = operand_is_id, |
| | 157 | // There may be unrelated decorations at the end, so make sure to |
| | 158 | // slice those off. |
| | 159 | .decorations = decorations.items(.entity)[0..decoration_i], |
| 110 | }; | 160 | }; |
| 111 | } | 161 | } |
| | 162 | |
| | 163 | fn entityDecorationsByIndex(self: ModuleInfo, index: usize) []const Entity { |
| | 164 | const values = self.entities.values(); |
| | 165 | const first_decoration = values[index].first_decoration; |
| | 166 | if (index == values.len - 1) { |
| | 167 | return self.decorations[first_decoration..]; |
| | 168 | } else { |
| | 169 | const next_first_decoration = values[index + 1].first_decoration; |
| | 170 | return self.decorations[first_decoration..next_first_decoration]; |
| | 171 | } |
| | 172 | } |
| 112 | }; | 173 | }; |
| 113 | | 174 | |
| 114 | const EntityContext = struct { | 175 | const EntityContext = struct { |
| ... | @@ -138,23 +199,39 @@ const EntityContext = struct { | ... | @@ -138,23 +199,39 @@ const EntityContext = struct { |
| 138 | return hasher.final(); | 199 | return hasher.final(); |
| 139 | } | 200 | } |
| 140 | | 201 | |
| 141 | fn hashInner(self: *EntityContext, hasher: *std.hash.Wyhash, id: ResultId) !void { | 202 | fn hashInner(self: *EntityContext, hasher: *std.hash.Wyhash, id: ResultId) error{OutOfMemory}!void { |
| 142 | const index = self.info.entities.getIndex(id).?; | 203 | const index = self.info.entities.getIndex(id) orelse { |
| | 204 | // Index unknown, the type or constant may depend on another result-id |
| | 205 | // that couldn't be deduplicated and so it wasn't added to info.entities. |
| | 206 | // In this case, just has the ID itself. |
| | 207 | std.hash.autoHash(hasher, id); |
| | 208 | return; |
| | 209 | }; |
| | 210 | |
| 143 | const entity = self.info.entities.values()[index]; | 211 | const entity = self.info.entities.values()[index]; |
| 144 | | 212 | |
| 145 | std.hash.autoHash(hasher, entity.kind); | | |
| 146 | if (entity.kind == .OpTypePointer) { | 213 | if (entity.kind == .OpTypePointer) { |
| 147 | // This may be either a pointer that is forward-referenced in the future, | 214 | // This may be either a pointer that is forward-referenced in the future, |
| 148 | // or a forward reference to a pointer. | 215 | // or a forward reference to a pointer. |
| 149 | const entry = try self.ptr_map_a.getOrPut(self.a, id); | 216 | const entry = try self.ptr_map_a.getOrPut(self.a, id); |
| 150 | if (entry.found_existing) { | 217 | if (entry.found_existing) { |
| 151 | // Pointer already seen. Hash the index instead of recursing into its children. | 218 | // Pointer already seen. Hash the index instead of recursing into its children. |
| 152 | // TODO: Discriminate this path somehow? | | |
| 153 | std.hash.autoHash(hasher, entry.index); | 219 | std.hash.autoHash(hasher, entry.index); |
| 154 | return; | 220 | return; |
| 155 | } | 221 | } |
| 156 | } | 222 | } |
| 157 | | 223 | |
| | 224 | try self.hashEntity(hasher, entity); |
| | 225 | |
| | 226 | // Process decorations. |
| | 227 | const decorations = self.info.entityDecorationsByIndex(index); |
| | 228 | for (decorations) |decoration| { |
| | 229 | try self.hashEntity(hasher, decoration); |
| | 230 | } |
| | 231 | } |
| | 232 | |
| | 233 | fn hashEntity(self: *EntityContext, hasher: *std.hash.Wyhash, entity: ModuleInfo.Entity) !void { |
| | 234 | std.hash.autoHash(hasher, entity.kind); |
| 158 | // Process operands | 235 | // Process operands |
| 159 | const operands = self.binary.instructions[entity.first_operand..][0..entity.num_operands]; | 236 | const operands = self.binary.instructions[entity.first_operand..][0..entity.num_operands]; |
| 160 | for (operands, 0..) |operand, i| { | 237 | for (operands, 0..) |operand, i| { |
| ... | @@ -178,19 +255,24 @@ const EntityContext = struct { | ... | @@ -178,19 +255,24 @@ const EntityContext = struct { |
| 178 | return try self.eqlInner(a, b); | 255 | return try self.eqlInner(a, b); |
| 179 | } | 256 | } |
| 180 | | 257 | |
| 181 | fn eqlInner(self: *EntityContext, id_a: ResultId, id_b: ResultId) !bool { | 258 | fn eqlInner(self: *EntityContext, id_a: ResultId, id_b: ResultId) error{OutOfMemory}!bool { |
| 182 | const index_a = self.info.entities.getIndex(id_a).?; | 259 | const maybe_index_a = self.info.entities.getIndex(id_a); |
| 183 | const index_b = self.info.entities.getIndex(id_b).?; | 260 | const maybe_index_b = self.info.entities.getIndex(id_b); |
| | 261 | |
| | 262 | if (maybe_index_a == null and maybe_index_b == null) { |
| | 263 | // Both indices unknown. In this case the type or constant |
| | 264 | // may depend on another result-id that couldn't be deduplicated |
| | 265 | // (so it wasn't added to info.entities). In this case, that particular |
| | 266 | // result-id should be the same one. |
| | 267 | return id_a == id_b; |
| | 268 | } |
| | 269 | |
| | 270 | const index_a = maybe_index_a orelse return false; |
| | 271 | const index_b = maybe_index_b orelse return false; |
| 184 | | 272 | |
| 185 | const entity_a = self.info.entities.values()[index_a]; | 273 | const entity_a = self.info.entities.values()[index_a]; |
| 186 | const entity_b = self.info.entities.values()[index_b]; | 274 | const entity_b = self.info.entities.values()[index_b]; |
| 187 | | 275 | |
| 188 | if (entity_a.kind != entity_b.kind) { | | |
| 189 | return false; | | |
| 190 | } else if (entity_a.result_id_index != entity_a.result_id_index) { | | |
| 191 | return false; | | |
| 192 | } | | |
| 193 | | | |
| 194 | if (entity_a.kind == .OpTypePointer) { | 276 | if (entity_a.kind == .OpTypePointer) { |
| 195 | // May be a forward reference, or should be saved as a potential | 277 | // May be a forward reference, or should be saved as a potential |
| 196 | // forward reference in the future. Whatever the case, it should | 278 | // forward reference in the future. Whatever the case, it should |
| ... | @@ -207,6 +289,33 @@ const EntityContext = struct { | ... | @@ -207,6 +289,33 @@ const EntityContext = struct { |
| 207 | } | 289 | } |
| 208 | } | 290 | } |
| 209 | | 291 | |
| | 292 | if (!try self.eqlEntities(entity_a, entity_b)) { |
| | 293 | return false; |
| | 294 | } |
| | 295 | |
| | 296 | // Compare decorations. |
| | 297 | const decorations_a = self.info.entityDecorationsByIndex(index_a); |
| | 298 | const decorations_b = self.info.entityDecorationsByIndex(index_b); |
| | 299 | if (decorations_a.len != decorations_b.len) { |
| | 300 | return false; |
| | 301 | } |
| | 302 | |
| | 303 | for (decorations_a, decorations_b) |decoration_a, decoration_b| { |
| | 304 | if (!try self.eqlEntities(decoration_a, decoration_b)) { |
| | 305 | return false; |
| | 306 | } |
| | 307 | } |
| | 308 | |
| | 309 | return true; |
| | 310 | } |
| | 311 | |
| | 312 | fn eqlEntities(self: *EntityContext, entity_a: ModuleInfo.Entity, entity_b: ModuleInfo.Entity) !bool { |
| | 313 | if (entity_a.kind != entity_b.kind) { |
| | 314 | return false; |
| | 315 | } else if (entity_a.result_id_index != entity_a.result_id_index) { |
| | 316 | return false; |
| | 317 | } |
| | 318 | |
| 210 | const operands_a = self.binary.instructions[entity_a.first_operand..][0..entity_a.num_operands]; | 319 | const operands_a = self.binary.instructions[entity_a.first_operand..][0..entity_a.num_operands]; |
| 211 | const operands_b = self.binary.instructions[entity_b.first_operand..][0..entity_b.num_operands]; | 320 | const operands_b = self.binary.instructions[entity_b.first_operand..][0..entity_b.num_operands]; |
| 212 | | 321 | |
| ... | @@ -260,7 +369,6 @@ pub fn run(parser: *BinaryModule.Parser, binary: *BinaryModule) !void { | ... | @@ -260,7 +369,6 @@ pub fn run(parser: *BinaryModule.Parser, binary: *BinaryModule) !void { |
| 260 | const a = arena.allocator(); | 369 | const a = arena.allocator(); |
| 261 | | 370 | |
| 262 | const info = try ModuleInfo.parse(a, parser, binary.*); | 371 | const info = try ModuleInfo.parse(a, parser, binary.*); |
| 263 | log.info("added {} entities", .{info.entities.count()}); | | |
| 264 | | 372 | |
| 265 | // Hash all keys once so that the maps can be allocated the right size. | 373 | // Hash all keys once so that the maps can be allocated the right size. |
| 266 | var ctx = EntityContext{ | 374 | var ctx = EntityContext{ |
| ... | @@ -280,10 +388,9 @@ pub fn run(parser: *BinaryModule.Parser, binary: *BinaryModule) !void { | ... | @@ -280,10 +388,9 @@ pub fn run(parser: *BinaryModule.Parser, binary: *BinaryModule) !void { |
| 280 | .entity_context = &ctx, | 388 | .entity_context = &ctx, |
| 281 | }); | 389 | }); |
| 282 | var replace = std.AutoArrayHashMap(ResultId, ResultId).init(a); | 390 | var replace = std.AutoArrayHashMap(ResultId, ResultId).init(a); |
| 283 | for (info.entities.keys(), info.entities.values()) |id, entity| { | 391 | for (info.entities.keys()) |id| { |
| 284 | const entry = try map.getOrPut(id); | 392 | const entry = try map.getOrPut(id); |
| 285 | if (entry.found_existing) { | 393 | if (entry.found_existing) { |
| 286 | log.info("deduplicating {} - {s} (prior definition: {})", .{ id, @tagName(entity.kind), entry.key_ptr.* }); | | |
| 287 | try replace.putNoClobber(id, entry.key_ptr.*); | 394 | try replace.putNoClobber(id, entry.key_ptr.*); |
| 288 | } | 395 | } |
| 289 | } | 396 | } |
| ... | @@ -297,13 +404,15 @@ pub fn run(parser: *BinaryModule.Parser, binary: *BinaryModule) !void { | ... | @@ -297,13 +404,15 @@ pub fn run(parser: *BinaryModule.Parser, binary: *BinaryModule) !void { |
| 297 | while (it.next()) |inst| { | 404 | while (it.next()) |inst| { |
| 298 | // Result-id can only be the first or second operand | 405 | // Result-id can only be the first or second operand |
| 299 | const inst_spec = parser.getInstSpec(inst.opcode).?; | 406 | const inst_spec = parser.getInstSpec(inst.opcode).?; |
| 300 | const maybe_result_id: ?ResultId = for (0..2) |i| { | 407 | |
| | 408 | const maybe_result_id_offset: ?u16 = for (0..2) |i| { |
| 301 | if (inst_spec.operands.len > i and inst_spec.operands[i].kind == .IdResult) { | 409 | if (inst_spec.operands.len > i and inst_spec.operands[i].kind == .IdResult) { |
| 302 | break @enumFromInt(inst.operands[i]); | 410 | break @intCast(i); |
| 303 | } | 411 | } |
| 304 | } else null; | 412 | } else null; |
| 305 | | 413 | |
| 306 | if (maybe_result_id) |result_id| { | 414 | if (maybe_result_id_offset) |offset| { |
| | 415 | const result_id: ResultId = @enumFromInt(inst.operands[offset]); |
| 307 | if (replace.contains(result_id)) continue; | 416 | if (replace.contains(result_id)) continue; |
| 308 | } | 417 | } |
| 309 | | 418 | |
| ... | @@ -312,8 +421,16 @@ pub fn run(parser: *BinaryModule.Parser, binary: *BinaryModule) !void { | ... | @@ -312,8 +421,16 @@ pub fn run(parser: *BinaryModule.Parser, binary: *BinaryModule) !void { |
| 312 | new_functions_section = section.instructions.items.len; | 421 | new_functions_section = section.instructions.items.len; |
| 313 | }, | 422 | }, |
| 314 | .OpTypeForwardPointer => continue, // We re-emit these where needed | 423 | .OpTypeForwardPointer => continue, // We re-emit these where needed |
| 315 | // TODO: These aren't supported yet, strip them out for testing purposes. | 424 | else => {}, |
| 316 | .OpName, .OpMemberName => continue, | 425 | } |
| | 426 | |
| | 427 | switch (inst.opcode.class()) { |
| | 428 | .Annotation, .Debug => { |
| | 429 | // For decoration-style instructions, only emit them |
| | 430 | // if the target is not removed. |
| | 431 | const target: ResultId = @enumFromInt(inst.operands[0]); |
| | 432 | if (replace.contains(target)) continue; |
| | 433 | }, |
| 317 | else => {}, | 434 | else => {}, |
| 318 | } | 435 | } |
| 319 | | 436 | |
| ... | @@ -330,9 +447,8 @@ pub fn run(parser: *BinaryModule.Parser, binary: *BinaryModule) !void { | ... | @@ -330,9 +447,8 @@ pub fn run(parser: *BinaryModule.Parser, binary: *BinaryModule) !void { |
| 330 | operand.* = @intFromEnum(new_id); | 447 | operand.* = @intFromEnum(new_id); |
| 331 | } | 448 | } |
| 332 | | 449 | |
| 333 | const id: ResultId = @enumFromInt(operand.*); | 450 | if (maybe_result_id_offset == null or maybe_result_id_offset.? != i) { |
| 334 | // TODO: This test is a little janky. Check the offset instead? | 451 | const id: ResultId = @enumFromInt(operand.*); |
| 335 | if (maybe_result_id == null or maybe_result_id.? != id) { | | |
| 336 | const index = info.entities.getIndex(id) orelse continue; | 452 | const index = info.entities.getIndex(id) orelse continue; |
| 337 | const entity = info.entities.values()[index]; | 453 | const entity = info.entities.values()[index]; |
| 338 | if (entity.kind == .OpTypePointer and !emitted_ptrs.contains(id)) { | 454 | if (entity.kind == .OpTypePointer and !emitted_ptrs.contains(id)) { |
| ... | @@ -349,7 +465,8 @@ pub fn run(parser: *BinaryModule.Parser, binary: *BinaryModule) !void { | ... | @@ -349,7 +465,8 @@ pub fn run(parser: *BinaryModule.Parser, binary: *BinaryModule) !void { |
| 349 | } | 465 | } |
| 350 | | 466 | |
| 351 | if (inst.opcode == .OpTypePointer) { | 467 | if (inst.opcode == .OpTypePointer) { |
| 352 | try emitted_ptrs.put(maybe_result_id.?, {}); | 468 | const result_id: ResultId = @enumFromInt(new_operands.items[maybe_result_id_offset.?]); |
| | 469 | try emitted_ptrs.put(result_id, {}); |
| 353 | } | 470 | } |
| 354 | | 471 | |
| 355 | try section.emitRawInstruction(a, inst.opcode, new_operands.items); | 472 | try section.emitRawInstruction(a, inst.opcode, new_operands.items); |
Robin Voetter <robin@voetter.nl> 2024-03-30 10:06:55+01:00
Signed by SSH key SHA256:ZS52FNyUv2WUXvO4njmVaFVO46RHojFuOrxRc4LuKzg
src