| author |  Andrew Kelley <andrew@ziglang.org> 2025-01-14 23:38:36-08:00 |
| committer | Andrew Kelley <andrew@ziglang.org> 2025-01-15 15:11:37-08:00 |
| log | ae1641412124a4b09c8274b18792c162bc3cf65b |
| tree | 3cd6eb62635f7a053381b9e53b2c2e87cc38a667 |
| parent | a7bd1a631b9dd5b1221fdc0f3e4b299bda18138c |
wasm linker: ability to get data and functions from objects
2 files changed, 106 insertions(+), 42 deletions(-)
src/link/Wasm.zig+43-6
| ... | @@ -180,8 +180,10 @@ dump_argv_list: std.ArrayListUnmanaged([]const u8), | ... | @@ -180,8 +180,10 @@ dump_argv_list: std.ArrayListUnmanaged([]const u8), |
| 180 | preloaded_strings: PreloadedStrings, | 180 | preloaded_strings: PreloadedStrings, |
| 181 | 181 | ||
| 182 | /// This field is used when emitting an object; `navs_exe` used otherwise. | 182 | /// This field is used when emitting an object; `navs_exe` used otherwise. |
| 183 | /// Does not include externs since that data lives elsewhere. | ||
| 183 | navs_obj: std.AutoArrayHashMapUnmanaged(InternPool.Nav.Index, ZcuDataObj) = .empty, | 184 | navs_obj: std.AutoArrayHashMapUnmanaged(InternPool.Nav.Index, ZcuDataObj) = .empty, |
| 184 | /// This field is unused when emitting an object; `navs_obj` used otherwise. | 185 | /// This field is unused when emitting an object; `navs_obj` used otherwise. |
| 186 | /// Does not include externs since that data lives elsewhere. | ||
| 185 | navs_exe: std.AutoArrayHashMapUnmanaged(InternPool.Nav.Index, ZcuDataExe) = .empty, | 187 | navs_exe: std.AutoArrayHashMapUnmanaged(InternPool.Nav.Index, ZcuDataExe) = .empty, |
| 186 | /// Tracks all InternPool values referenced by codegen. Needed for outputting | 188 | /// Tracks all InternPool values referenced by codegen. Needed for outputting |
| 187 | /// the data segment. This one does not track ref count because object files | 189 | /// the data segment. This one does not track ref count because object files |
| ... | @@ -221,6 +223,9 @@ functions: std.AutoArrayHashMapUnmanaged(FunctionImport.Resolution, void) = .emp | ... | @@ -221,6 +223,9 @@ functions: std.AutoArrayHashMapUnmanaged(FunctionImport.Resolution, void) = .emp |
| 221 | /// Tracks the value at the end of prelink, at which point `functions` | 223 | /// Tracks the value at the end of prelink, at which point `functions` |
| 222 | /// contains only object file functions, and nothing from the Zcu yet. | 224 | /// contains only object file functions, and nothing from the Zcu yet. |
| 223 | functions_end_prelink: u32 = 0, | 225 | functions_end_prelink: u32 = 0, |
| 226 | |||
| 227 | function_imports_len_prelink: u32 = 0, | ||
| 228 | data_imports_len_prelink: u32 = 0, | ||
| 224 | /// At the end of prelink, this is populated with needed functions from | 229 | /// At the end of prelink, this is populated with needed functions from |
| 225 | /// objects. | 230 | /// objects. |
| 226 | /// | 231 | /// |
| ... | @@ -3447,6 +3452,9 @@ pub fn prelink(wasm: *Wasm, prog_node: std.Progress.Node) link.File.FlushError!v | ... | @@ -3447,6 +3452,9 @@ pub fn prelink(wasm: *Wasm, prog_node: std.Progress.Node) link.File.FlushError!v |
| 3447 | wasm.memories.limits.max = @max(wasm.memories.limits.max, memory_import.limits_max); | 3452 | wasm.memories.limits.max = @max(wasm.memories.limits.max, memory_import.limits_max); |
| 3448 | wasm.memories.limits.flags.has_max = wasm.memories.limits.flags.has_max or memory_import.limits_has_max; | 3453 | wasm.memories.limits.flags.has_max = wasm.memories.limits.flags.has_max or memory_import.limits_has_max; |
| 3449 | } | 3454 | } |
| 3455 | |||
| 3456 | wasm.function_imports_len_prelink = @intCast(wasm.function_imports.entries.len); | ||
| 3457 | wasm.data_imports_len_prelink = @intCast(wasm.data_imports.entries.len); | ||
| 3450 | } | 3458 | } |
| 3451 | 3459 | ||
| 3452 | pub fn markFunctionImport( | 3460 | pub fn markFunctionImport( |
| ... | @@ -3608,7 +3616,7 @@ fn markDataSegment(wasm: *Wasm, segment_index: ObjectDataSegment.Index) link.Fil | ... | @@ -3608,7 +3616,7 @@ fn markDataSegment(wasm: *Wasm, segment_index: ObjectDataSegment.Index) link.Fil |
| 3608 | try wasm.markRelocations(segment.relocations(wasm)); | 3616 | try wasm.markRelocations(segment.relocations(wasm)); |
| 3609 | } | 3617 | } |
| 3610 | 3618 | ||
| 3611 | fn markDataImport( | 3619 | pub fn markDataImport( |
| 3612 | wasm: *Wasm, | 3620 | wasm: *Wasm, |
| 3613 | name: String, | 3621 | name: String, |
| 3614 | import: *ObjectDataImport, | 3622 | import: *ObjectDataImport, |
| ... | @@ -4499,11 +4507,40 @@ pub fn navAddr(wasm: *Wasm, nav_index: InternPool.Nav.Index) u32 { | ... | @@ -4499,11 +4507,40 @@ pub fn navAddr(wasm: *Wasm, nav_index: InternPool.Nav.Index) u32 { |
| 4499 | assert(wasm.flush_buffer.memory_layout_finished); | 4507 | assert(wasm.flush_buffer.memory_layout_finished); |
| 4500 | const comp = wasm.base.comp; | 4508 | const comp = wasm.base.comp; |
| 4501 | assert(comp.config.output_mode != .Obj); | 4509 | assert(comp.config.output_mode != .Obj); |
| 4502 | // If there is no entry it means the value is zero bits so any address will do. | 4510 | if (wasm.navs_exe.getIndex(nav_index)) |i| { |
| 4503 | const navs_exe_index: NavsExeIndex = @enumFromInt(wasm.navs_exe.getIndex(nav_index) orelse return 0); | 4511 | const navs_exe_index: NavsExeIndex = @enumFromInt(i); |
| 4504 | log.debug("navAddr {s} {}", .{ navs_exe_index.name(wasm), nav_index }); | 4512 | log.debug("navAddr {s} {}", .{ navs_exe_index.name(wasm), nav_index }); |
| 4505 | const ds_id: DataSegmentId = .pack(wasm, .{ .nav_exe = navs_exe_index }); | 4513 | const ds_id: DataSegmentId = .pack(wasm, .{ .nav_exe = navs_exe_index }); |
| 4506 | return wasm.flush_buffer.data_segments.get(ds_id).?; | 4514 | return wasm.flush_buffer.data_segments.get(ds_id).?; |
| 4515 | } | ||
| 4516 | const zcu = comp.zcu.?; | ||
| 4517 | const ip = &zcu.intern_pool; | ||
| 4518 | const nav = ip.getNav(nav_index); | ||
| 4519 | if (nav.getResolvedExtern(ip)) |ext| { | ||
| 4520 | if (wasm.getExistingString(ext.name.toSlice(ip))) |symbol_name| { | ||
| 4521 | if (wasm.object_data_imports.getPtr(symbol_name)) |import| { | ||
| 4522 | switch (import.resolution.unpack(wasm)) { | ||
| 4523 | .unresolved => unreachable, | ||
| 4524 | .object => |object_data_index| { | ||
| 4525 | const object_data = object_data_index.ptr(wasm); | ||
| 4526 | const ds_id: DataSegmentId = .fromObjectDataSegment(wasm, object_data.segment); | ||
| 4527 | const segment_base_addr = wasm.flush_buffer.data_segments.get(ds_id).?; | ||
| 4528 | return segment_base_addr + object_data.offset; | ||
| 4529 | }, | ||
| 4530 | .__zig_error_names => @panic("TODO"), | ||
| 4531 | .__zig_error_name_table => @panic("TODO"), | ||
| 4532 | .__heap_base => @panic("TODO"), | ||
| 4533 | .__heap_end => @panic("TODO"), | ||
| 4534 | .uav_exe => @panic("TODO"), | ||
| 4535 | .uav_obj => @panic("TODO"), | ||
| 4536 | .nav_exe => @panic("TODO"), | ||
| 4537 | .nav_obj => @panic("TODO"), | ||
| 4538 | } | ||
| 4539 | } | ||
| 4540 | } | ||
| 4541 | } | ||
| 4542 | // Otherwise it's a zero bit type; any address will do. | ||
| 4543 | return 0; | ||
| 4507 | } | 4544 | } |
| 4508 | 4545 | ||
| 4509 | /// Asserts it is called after `Flush.data_segments` is fully populated and sorted. | 4546 | /// Asserts it is called after `Flush.data_segments` is fully populated and sorted. |
src/link/Wasm/Flush.zig+63-36
| ... | @@ -122,45 +122,71 @@ pub fn finish(f: *Flush, wasm: *Wasm) !void { | ... | @@ -122,45 +122,71 @@ pub fn finish(f: *Flush, wasm: *Wasm) !void { |
| 122 | 122 | ||
| 123 | const entry_name = if (wasm.entry_resolution.isNavOrUnresolved(wasm)) wasm.entry_name else .none; | 123 | const entry_name = if (wasm.entry_resolution.isNavOrUnresolved(wasm)) wasm.entry_name else .none; |
| 124 | 124 | ||
| 125 | // Detect any intrinsics that were called; they need to have dependencies on the symbols marked. | 125 | if (comp.zcu) |zcu| { |
| 126 | // Likewise detect `@tagName` calls so those functions can be included in the output and synthesized. | 126 | const ip: *const InternPool = &zcu.intern_pool; // No mutations allowed! |
| 127 | for (wasm.mir_instructions.items(.tag), wasm.mir_instructions.items(.data)) |tag, *data| switch (tag) { | 127 | |
| 128 | .call_intrinsic => { | 128 | // Detect any intrinsics that were called; they need to have dependencies on the symbols marked. |
| 129 | const symbol_name = try wasm.internString(@tagName(data.intrinsic)); | 129 | // Likewise detect `@tagName` calls so those functions can be included in the output and synthesized. |
| 130 | const i: Wasm.FunctionImport.Index = @enumFromInt(wasm.object_function_imports.getIndex(symbol_name) orelse { | 130 | for (wasm.mir_instructions.items(.tag), wasm.mir_instructions.items(.data)) |tag, *data| switch (tag) { |
| 131 | return diags.fail("missing compiler runtime intrinsic '{s}' (undefined linker symbol)", .{ | 131 | .call_intrinsic => { |
| 132 | @tagName(data.intrinsic), | 132 | const symbol_name = try wasm.internString(@tagName(data.intrinsic)); |
| 133 | const i: Wasm.FunctionImport.Index = @enumFromInt(wasm.object_function_imports.getIndex(symbol_name) orelse { | ||
| 134 | return diags.fail("missing compiler runtime intrinsic '{s}' (undefined linker symbol)", .{ | ||
| 135 | @tagName(data.intrinsic), | ||
| 136 | }); | ||
| 133 | }); | 137 | }); |
| 134 | }); | 138 | try wasm.markFunctionImport(symbol_name, i.value(wasm), i); |
| 135 | try wasm.markFunctionImport(symbol_name, i.value(wasm), i); | 139 | }, |
| 136 | }, | 140 | .call_tag_name => { |
| 137 | .call_tag_name => { | 141 | assert(ip.indexToKey(data.ip_index) == .enum_type); |
| 138 | const zcu = comp.zcu.?; | 142 | const gop = try wasm.zcu_funcs.getOrPut(gpa, data.ip_index); |
| 139 | const ip = &zcu.intern_pool; | 143 | if (!gop.found_existing) { |
| 140 | assert(ip.indexToKey(data.ip_index) == .enum_type); | 144 | wasm.tag_name_table_ref_count += 1; |
| 141 | const gop = try wasm.zcu_funcs.getOrPut(gpa, data.ip_index); | 145 | const int_tag_ty = Zcu.Type.fromInterned(data.ip_index).intTagType(zcu); |
| 142 | if (!gop.found_existing) { | 146 | gop.value_ptr.* = .{ .tag_name = .{ |
| 143 | wasm.tag_name_table_ref_count += 1; | 147 | .symbol_name = try wasm.internStringFmt("__zig_tag_name_{d}", .{@intFromEnum(data.ip_index)}), |
| 144 | const int_tag_ty = Zcu.Type.fromInterned(data.ip_index).intTagType(zcu); | 148 | .type_index = try wasm.internFunctionType(.Unspecified, &.{int_tag_ty.ip_index}, .slice_const_u8_sentinel_0, target), |
| 145 | gop.value_ptr.* = .{ .tag_name = .{ | 149 | .table_index = @intCast(wasm.tag_name_offs.items.len), |
| 146 | .symbol_name = try wasm.internStringFmt("__zig_tag_name_{d}", .{@intFromEnum(data.ip_index)}), | 150 | } }; |
| 147 | .type_index = try wasm.internFunctionType(.Unspecified, &.{int_tag_ty.ip_index}, .slice_const_u8_sentinel_0, target), | 151 | try wasm.functions.put(gpa, .fromZcuFunc(wasm, @enumFromInt(gop.index)), {}); |
| 148 | .table_index = @intCast(wasm.tag_name_offs.items.len), | 152 | const tag_names = ip.loadEnumType(data.ip_index).names; |
| 149 | } }; | 153 | for (tag_names.get(ip)) |tag_name| { |
| 150 | try wasm.functions.put(gpa, .fromZcuFunc(wasm, @enumFromInt(gop.index)), {}); | 154 | const slice = tag_name.toSlice(ip); |
| 151 | const tag_names = ip.loadEnumType(data.ip_index).names; | 155 | try wasm.tag_name_offs.append(gpa, @intCast(wasm.tag_name_bytes.items.len)); |
| 152 | for (tag_names.get(ip)) |tag_name| { | 156 | try wasm.tag_name_bytes.appendSlice(gpa, slice[0 .. slice.len + 1]); |
| 153 | const slice = tag_name.toSlice(ip); | 157 | } |
| 154 | try wasm.tag_name_offs.append(gpa, @intCast(wasm.tag_name_bytes.items.len)); | ||
| 155 | try wasm.tag_name_bytes.appendSlice(gpa, slice[0 .. slice.len + 1]); | ||
| 156 | } | 158 | } |
| 159 | }, | ||
| 160 | else => continue, | ||
| 161 | }; | ||
| 162 | |||
| 163 | { | ||
| 164 | var i = wasm.function_imports_len_prelink; | ||
| 165 | while (i < f.function_imports.entries.len) { | ||
| 166 | const symbol_name = f.function_imports.keys()[i]; | ||
| 167 | if (wasm.object_function_imports.getIndex(symbol_name)) |import_index_usize| { | ||
| 168 | const import_index: Wasm.FunctionImport.Index = @enumFromInt(import_index_usize); | ||
| 169 | try wasm.markFunctionImport(symbol_name, import_index.value(wasm), import_index); | ||
| 170 | f.function_imports.swapRemoveAt(i); | ||
| 171 | continue; | ||
| 172 | } | ||
| 173 | i += 1; | ||
| 157 | } | 174 | } |
| 158 | }, | 175 | } |
| 159 | else => continue, | ||
| 160 | }; | ||
| 161 | 176 | ||
| 162 | if (comp.zcu) |zcu| { | 177 | { |
| 163 | const ip: *const InternPool = &zcu.intern_pool; // No mutations allowed! | 178 | var i = wasm.data_imports_len_prelink; |
| 179 | while (i < f.data_imports.entries.len) { | ||
| 180 | const symbol_name = f.data_imports.keys()[i]; | ||
| 181 | if (wasm.object_data_imports.getIndex(symbol_name)) |import_index_usize| { | ||
| 182 | const import_index: Wasm.ObjectDataImport.Index = @enumFromInt(import_index_usize); | ||
| 183 | try wasm.markDataImport(symbol_name, import_index.value(wasm), import_index); | ||
| 184 | f.data_imports.swapRemoveAt(i); | ||
| 185 | continue; | ||
| 186 | } | ||
| 187 | i += 1; | ||
| 188 | } | ||
| 189 | } | ||
| 164 | 190 | ||
| 165 | if (wasm.error_name_table_ref_count > 0) { | 191 | if (wasm.error_name_table_ref_count > 0) { |
| 166 | // Ensure Zcu error name structures are populated. | 192 | // Ensure Zcu error name structures are populated. |
| ... | @@ -437,7 +463,7 @@ pub fn finish(f: *Flush, wasm: *Wasm) !void { | ... | @@ -437,7 +463,7 @@ pub fn finish(f: *Flush, wasm: *Wasm) !void { |
| 437 | break :b i >= 1 and !wantSegmentMerge(wasm, segment_ids[i - 1], segment_id, category); | 463 | break :b i >= 1 and !wantSegmentMerge(wasm, segment_ids[i - 1], segment_id, category); |
| 438 | }; | 464 | }; |
| 439 | if (want_new_segment) { | 465 | if (want_new_segment) { |
| 440 | log.debug("new segment at 0x{x} {} {s} {}", .{ start_addr, segment_id, segment_id.name(wasm), category }); | 466 | log.debug("new segment group at 0x{x} {} {s} {}", .{ start_addr, segment_id, segment_id.name(wasm), category }); |
| 441 | try f.data_segment_groups.append(gpa, .{ | 467 | try f.data_segment_groups.append(gpa, .{ |
| 442 | .end_addr = @intCast(memory_ptr), | 468 | .end_addr = @intCast(memory_ptr), |
| 443 | .first_segment = first_segment, | 469 | .first_segment = first_segment, |
| ... | @@ -447,6 +473,7 @@ pub fn finish(f: *Flush, wasm: *Wasm) !void { | ... | @@ -447,6 +473,7 @@ pub fn finish(f: *Flush, wasm: *Wasm) !void { |
| 447 | 473 | ||
| 448 | const size = segment_id.size(wasm); | 474 | const size = segment_id.size(wasm); |
| 449 | segment_vaddr.* = @intCast(start_addr); | 475 | segment_vaddr.* = @intCast(start_addr); |
| 476 | log.debug("0x{x} {d} {s}", .{ start_addr, @intFromEnum(segment_id), segment_id.name(wasm) }); | ||
| 450 | memory_ptr = start_addr + size; | 477 | memory_ptr = start_addr + size; |
| 451 | } | 478 | } |
| 452 | if (category != .zero) try f.data_segment_groups.append(gpa, .{ | 479 | if (category != .zero) try f.data_segment_groups.append(gpa, .{ |