wasm linker: ability to get data and functions from objects

This commit is contained in:
Andrew Kelley 2025-01-14 23:38:36 -08:00
parent a7bd1a631b
commit ae16414121
2 changed files with 108 additions and 44 deletions

View File

@ -180,8 +180,10 @@ dump_argv_list: std.ArrayListUnmanaged([]const u8),
preloaded_strings: PreloadedStrings,
/// This field is used when emitting an object; `navs_exe` used otherwise.
/// Does not include externs since that data lives elsewhere.
navs_obj: std.AutoArrayHashMapUnmanaged(InternPool.Nav.Index, ZcuDataObj) = .empty,
/// This field is unused when emitting an object; `navs_obj` used otherwise.
/// Does not include externs since that data lives elsewhere.
navs_exe: std.AutoArrayHashMapUnmanaged(InternPool.Nav.Index, ZcuDataExe) = .empty,
/// Tracks all InternPool values referenced by codegen. Needed for outputting
/// the data segment. This one does not track ref count because object files
@ -221,6 +223,9 @@ functions: std.AutoArrayHashMapUnmanaged(FunctionImport.Resolution, void) = .emp
/// Tracks the value at the end of prelink, at which point `functions`
/// contains only object file functions, and nothing from the Zcu yet.
functions_end_prelink: u32 = 0,
function_imports_len_prelink: u32 = 0,
data_imports_len_prelink: u32 = 0,
/// At the end of prelink, this is populated with needed functions from
/// objects.
///
@ -3447,6 +3452,9 @@ pub fn prelink(wasm: *Wasm, prog_node: std.Progress.Node) link.File.FlushError!v
wasm.memories.limits.max = @max(wasm.memories.limits.max, memory_import.limits_max);
wasm.memories.limits.flags.has_max = wasm.memories.limits.flags.has_max or memory_import.limits_has_max;
}
wasm.function_imports_len_prelink = @intCast(wasm.function_imports.entries.len);
wasm.data_imports_len_prelink = @intCast(wasm.data_imports.entries.len);
}
pub fn markFunctionImport(
@ -3608,7 +3616,7 @@ fn markDataSegment(wasm: *Wasm, segment_index: ObjectDataSegment.Index) link.Fil
try wasm.markRelocations(segment.relocations(wasm));
}
fn markDataImport(
pub fn markDataImport(
wasm: *Wasm,
name: String,
import: *ObjectDataImport,
@ -4499,11 +4507,40 @@ pub fn navAddr(wasm: *Wasm, nav_index: InternPool.Nav.Index) u32 {
assert(wasm.flush_buffer.memory_layout_finished);
const comp = wasm.base.comp;
assert(comp.config.output_mode != .Obj);
// If there is no entry it means the value is zero bits so any address will do.
const navs_exe_index: NavsExeIndex = @enumFromInt(wasm.navs_exe.getIndex(nav_index) orelse return 0);
log.debug("navAddr {s} {}", .{ navs_exe_index.name(wasm), nav_index });
const ds_id: DataSegmentId = .pack(wasm, .{ .nav_exe = navs_exe_index });
return wasm.flush_buffer.data_segments.get(ds_id).?;
if (wasm.navs_exe.getIndex(nav_index)) |i| {
const navs_exe_index: NavsExeIndex = @enumFromInt(i);
log.debug("navAddr {s} {}", .{ navs_exe_index.name(wasm), nav_index });
const ds_id: DataSegmentId = .pack(wasm, .{ .nav_exe = navs_exe_index });
return wasm.flush_buffer.data_segments.get(ds_id).?;
}
const zcu = comp.zcu.?;
const ip = &zcu.intern_pool;
const nav = ip.getNav(nav_index);
if (nav.getResolvedExtern(ip)) |ext| {
if (wasm.getExistingString(ext.name.toSlice(ip))) |symbol_name| {
if (wasm.object_data_imports.getPtr(symbol_name)) |import| {
switch (import.resolution.unpack(wasm)) {
.unresolved => unreachable,
.object => |object_data_index| {
const object_data = object_data_index.ptr(wasm);
const ds_id: DataSegmentId = .fromObjectDataSegment(wasm, object_data.segment);
const segment_base_addr = wasm.flush_buffer.data_segments.get(ds_id).?;
return segment_base_addr + object_data.offset;
},
.__zig_error_names => @panic("TODO"),
.__zig_error_name_table => @panic("TODO"),
.__heap_base => @panic("TODO"),
.__heap_end => @panic("TODO"),
.uav_exe => @panic("TODO"),
.uav_obj => @panic("TODO"),
.nav_exe => @panic("TODO"),
.nav_obj => @panic("TODO"),
}
}
}
}
// Otherwise it's a zero bit type; any address will do.
return 0;
}
/// Asserts it is called after `Flush.data_segments` is fully populated and sorted.

View File

@ -122,46 +122,72 @@ pub fn finish(f: *Flush, wasm: *Wasm) !void {
const entry_name = if (wasm.entry_resolution.isNavOrUnresolved(wasm)) wasm.entry_name else .none;
// Detect any intrinsics that were called; they need to have dependencies on the symbols marked.
// Likewise detect `@tagName` calls so those functions can be included in the output and synthesized.
for (wasm.mir_instructions.items(.tag), wasm.mir_instructions.items(.data)) |tag, *data| switch (tag) {
.call_intrinsic => {
const symbol_name = try wasm.internString(@tagName(data.intrinsic));
const i: Wasm.FunctionImport.Index = @enumFromInt(wasm.object_function_imports.getIndex(symbol_name) orelse {
return diags.fail("missing compiler runtime intrinsic '{s}' (undefined linker symbol)", .{
@tagName(data.intrinsic),
});
});
try wasm.markFunctionImport(symbol_name, i.value(wasm), i);
},
.call_tag_name => {
const zcu = comp.zcu.?;
const ip = &zcu.intern_pool;
assert(ip.indexToKey(data.ip_index) == .enum_type);
const gop = try wasm.zcu_funcs.getOrPut(gpa, data.ip_index);
if (!gop.found_existing) {
wasm.tag_name_table_ref_count += 1;
const int_tag_ty = Zcu.Type.fromInterned(data.ip_index).intTagType(zcu);
gop.value_ptr.* = .{ .tag_name = .{
.symbol_name = try wasm.internStringFmt("__zig_tag_name_{d}", .{@intFromEnum(data.ip_index)}),
.type_index = try wasm.internFunctionType(.Unspecified, &.{int_tag_ty.ip_index}, .slice_const_u8_sentinel_0, target),
.table_index = @intCast(wasm.tag_name_offs.items.len),
} };
try wasm.functions.put(gpa, .fromZcuFunc(wasm, @enumFromInt(gop.index)), {});
const tag_names = ip.loadEnumType(data.ip_index).names;
for (tag_names.get(ip)) |tag_name| {
const slice = tag_name.toSlice(ip);
try wasm.tag_name_offs.append(gpa, @intCast(wasm.tag_name_bytes.items.len));
try wasm.tag_name_bytes.appendSlice(gpa, slice[0 .. slice.len + 1]);
}
}
},
else => continue,
};
if (comp.zcu) |zcu| {
const ip: *const InternPool = &zcu.intern_pool; // No mutations allowed!
// Detect any intrinsics that were called; they need to have dependencies on the symbols marked.
// Likewise detect `@tagName` calls so those functions can be included in the output and synthesized.
for (wasm.mir_instructions.items(.tag), wasm.mir_instructions.items(.data)) |tag, *data| switch (tag) {
.call_intrinsic => {
const symbol_name = try wasm.internString(@tagName(data.intrinsic));
const i: Wasm.FunctionImport.Index = @enumFromInt(wasm.object_function_imports.getIndex(symbol_name) orelse {
return diags.fail("missing compiler runtime intrinsic '{s}' (undefined linker symbol)", .{
@tagName(data.intrinsic),
});
});
try wasm.markFunctionImport(symbol_name, i.value(wasm), i);
},
.call_tag_name => {
assert(ip.indexToKey(data.ip_index) == .enum_type);
const gop = try wasm.zcu_funcs.getOrPut(gpa, data.ip_index);
if (!gop.found_existing) {
wasm.tag_name_table_ref_count += 1;
const int_tag_ty = Zcu.Type.fromInterned(data.ip_index).intTagType(zcu);
gop.value_ptr.* = .{ .tag_name = .{
.symbol_name = try wasm.internStringFmt("__zig_tag_name_{d}", .{@intFromEnum(data.ip_index)}),
.type_index = try wasm.internFunctionType(.Unspecified, &.{int_tag_ty.ip_index}, .slice_const_u8_sentinel_0, target),
.table_index = @intCast(wasm.tag_name_offs.items.len),
} };
try wasm.functions.put(gpa, .fromZcuFunc(wasm, @enumFromInt(gop.index)), {});
const tag_names = ip.loadEnumType(data.ip_index).names;
for (tag_names.get(ip)) |tag_name| {
const slice = tag_name.toSlice(ip);
try wasm.tag_name_offs.append(gpa, @intCast(wasm.tag_name_bytes.items.len));
try wasm.tag_name_bytes.appendSlice(gpa, slice[0 .. slice.len + 1]);
}
}
},
else => continue,
};
{
var i = wasm.function_imports_len_prelink;
while (i < f.function_imports.entries.len) {
const symbol_name = f.function_imports.keys()[i];
if (wasm.object_function_imports.getIndex(symbol_name)) |import_index_usize| {
const import_index: Wasm.FunctionImport.Index = @enumFromInt(import_index_usize);
try wasm.markFunctionImport(symbol_name, import_index.value(wasm), import_index);
f.function_imports.swapRemoveAt(i);
continue;
}
i += 1;
}
}
{
var i = wasm.data_imports_len_prelink;
while (i < f.data_imports.entries.len) {
const symbol_name = f.data_imports.keys()[i];
if (wasm.object_data_imports.getIndex(symbol_name)) |import_index_usize| {
const import_index: Wasm.ObjectDataImport.Index = @enumFromInt(import_index_usize);
try wasm.markDataImport(symbol_name, import_index.value(wasm), import_index);
f.data_imports.swapRemoveAt(i);
continue;
}
i += 1;
}
}
if (wasm.error_name_table_ref_count > 0) {
// Ensure Zcu error name structures are populated.
const full_error_names = ip.global_error_set.getNamesFromMainThread();
@ -437,7 +463,7 @@ pub fn finish(f: *Flush, wasm: *Wasm) !void {
break :b i >= 1 and !wantSegmentMerge(wasm, segment_ids[i - 1], segment_id, category);
};
if (want_new_segment) {
log.debug("new segment at 0x{x} {} {s} {}", .{ start_addr, segment_id, segment_id.name(wasm), category });
log.debug("new segment group at 0x{x} {} {s} {}", .{ start_addr, segment_id, segment_id.name(wasm), category });
try f.data_segment_groups.append(gpa, .{
.end_addr = @intCast(memory_ptr),
.first_segment = first_segment,
@ -447,6 +473,7 @@ pub fn finish(f: *Flush, wasm: *Wasm) !void {
const size = segment_id.size(wasm);
segment_vaddr.* = @intCast(start_addr);
log.debug("0x{x} {d} {s}", .{ start_addr, @intFromEnum(segment_id), segment_id.name(wasm) });
memory_ptr = start_addr + size;
}
if (category != .zero) try f.data_segment_groups.append(gpa, .{