Merge pull request #13389 from jacobly0/fix-only-c

cbe: enough fixes for `-Donly-c` to be able to produce an executable

Author: andrewrk

build.zig

@@ -17,7 +17,7 @@ pub fn build(b: *Builder) !void {
     b.setPreferredReleaseMode(.ReleaseFast);
     const test_step = b.step("test", "Run all the tests");
     const mode = b.standardReleaseOptions();
-    const target = b.standardTargetOptions(.{});
+    var target = b.standardTargetOptions(.{});
     const single_threaded = b.option(bool, "single-threaded", "Build artifacts that run in single threaded mode");
     const use_zig_libcxx = b.option(bool, "use-zig-libcxx", "If libc++ is needed, use zig's bundled version, don't try to integrate with the system") orelse false;
 
@@ -141,6 +141,10 @@ pub fn build(b: *Builder) !void {
         break :blk 4;
     };
 
+    if (only_c) {
+        target.ofmt = .c;
+    }
+
     const main_file: ?[]const u8 = mf: {
         if (!have_stage1) break :mf "src/main.zig";
         if (use_zig0) break :mf null;
@@ -172,10 +176,6 @@ pub fn build(b: *Builder) !void {
         test_cases.want_lto = false;
     }
 
-    if (only_c) {
-        exe.ofmt = .c;
-    }
-
     const exe_options = b.addOptions();
     exe.addOptions("build_options", exe_options);
 

lib/include/zig.h

@@ -19,7 +19,7 @@
 #endif
 
 #if __STDC_VERSION__ >= 201112L
-#define zig_threadlocal thread_local
+#define zig_threadlocal _Thread_local
 #elif defined(__GNUC__)
 #define zig_threadlocal __thread
 #elif _MSC_VER

lib/std/build.zig

@@ -1622,7 +1622,6 @@ pub const LibExeObjStep = struct {
     use_stage1: ?bool = null,
     use_llvm: ?bool = null,
     use_lld: ?bool = null,
-    ofmt: ?std.Target.ObjectFormat = null,
 
     output_path_source: GeneratedFile,
     output_lib_path_source: GeneratedFile,
@@ -2490,7 +2489,7 @@ pub const LibExeObjStep = struct {
             }
         }
 
-        if (self.ofmt) |ofmt| {
+        if (self.target.ofmt) |ofmt| {
             try zig_args.append(try std.fmt.allocPrint(builder.allocator, "-ofmt={s}", .{@tagName(ofmt)}));
         }
 

lib/std/crypto/blake3.zig

@@ -200,7 +200,10 @@ const CompressGeneric = struct {
     }
 };
 
-const compress = if (builtin.cpu.arch == .x86_64) CompressVectorized.compress else CompressGeneric.compress;
+const compress = if (builtin.cpu.arch == .x86_64 and builtin.zig_backend != .stage2_c)
+    CompressVectorized.compress
+else
+    CompressGeneric.compress;
 
 fn first8Words(words: [16]u32) [8]u32 {
     return @ptrCast(*const [8]u32, &words).*;

lib/std/crypto/gimli.zig

@@ -152,7 +152,7 @@ pub const State = struct {
         self.endianSwap();
     }
 
-    pub const permute = if (builtin.cpu.arch == .x86_64) impl: {
+    pub const permute = if (builtin.cpu.arch == .x86_64 and builtin.zig_backend != .stage2_c) impl: {
         break :impl permute_vectorized;
     } else if (builtin.mode == .ReleaseSmall) impl: {
         break :impl permute_small;

lib/std/multi_array_list.zig

@@ -436,9 +436,15 @@ pub fn MultiArrayList(comptime S: type) type {
         }
 
         fn capacityInBytes(capacity: usize) usize {
-            const sizes_vector: @Vector(sizes.bytes.len, usize) = sizes.bytes;
-            const capacity_vector = @splat(sizes.bytes.len, capacity);
-            return @reduce(.Add, capacity_vector * sizes_vector);
+            if (builtin.zig_backend == .stage2_c) {
+                var bytes: usize = 0;
+                for (sizes.bytes) |size| bytes += size * capacity;
+                return bytes;
+            } else {
+                const sizes_vector: @Vector(sizes.bytes.len, usize) = sizes.bytes;
+                const capacity_vector = @splat(sizes.bytes.len, capacity);
+                return @reduce(.Add, capacity_vector * sizes_vector);
+            }
         }
 
         fn allocatedBytes(self: Self) []align(@alignOf(S)) u8 {

lib/std/target.zig

@@ -1,4 +1,5 @@
 const std = @import("std.zig");
+const builtin = @import("builtin");
 const mem = std.mem;
 const Version = std.builtin.Version;
 
@@ -719,7 +720,11 @@ pub const Target = struct {
 
                 /// Adds the specified feature set but not its dependencies.
                 pub fn addFeatureSet(set: *Set, other_set: Set) void {
-                    set.ints = @as(@Vector(usize_count, usize), set.ints) | @as(@Vector(usize_count, usize), other_set.ints);
+                    if (builtin.zig_backend == .stage2_c) {
+                        for (set.ints) |*int, i| int.* |= other_set.ints[i];
+                    } else {
+                        set.ints = @as(@Vector(usize_count, usize), set.ints) | @as(@Vector(usize_count, usize), other_set.ints);
+                    }
                 }
 
                 /// Removes the specified feature but not its dependents.
@@ -731,7 +736,11 @@ pub const Target = struct {
 
                 /// Removes the specified feature but not its dependents.
                 pub fn removeFeatureSet(set: *Set, other_set: Set) void {
-                    set.ints = @as(@Vector(usize_count, usize), set.ints) & ~@as(@Vector(usize_count, usize), other_set.ints);
+                    if (builtin.zig_backend == .stage2_c) {
+                        for (set.ints) |*int, i| int.* &= ~other_set.ints[i];
+                    } else {
+                        set.ints = @as(@Vector(usize_count, usize), set.ints) & ~@as(@Vector(usize_count, usize), other_set.ints);
+                    }
                 }
 
                 pub fn populateDependencies(set: *Set, all_features_list: []const Cpu.Feature) void {

lib/std/zig/system/x86.zig

@@ -528,35 +528,29 @@ const CpuidLeaf = packed struct {
 };
 
 fn cpuid(leaf_id: u32, subid: u32) CpuidLeaf {
-    // Workaround for https://github.com/ziglang/zig/issues/215
-    // Inline assembly in zig only supports one output,
-    // so we pass a pointer to the struct.
-    var cpuid_leaf: CpuidLeaf = undefined;
-
     // valid for both x86 and x86_64
-    asm volatile (
-        \\ cpuid
-        \\ movl %%eax, 0(%[leaf_ptr])
-        \\ movl %%ebx, 4(%[leaf_ptr])
-        \\ movl %%ecx, 8(%[leaf_ptr])
-        \\ movl %%edx, 12(%[leaf_ptr])
-        :
-        : [leaf_id] "{eax}" (leaf_id),
-          [subid] "{ecx}" (subid),
-          [leaf_ptr] "r" (&cpuid_leaf),
-        : "eax", "ebx", "ecx", "edx"
+    var eax: u32 = undefined;
+    var ebx: u32 = undefined;
+    var ecx: u32 = undefined;
+    var edx: u32 = undefined;
+    asm volatile ("cpuid"
+        : [_] "={eax}" (eax),
+          [_] "={ebx}" (ebx),
+          [_] "={ecx}" (ecx),
+          [_] "={edx}" (edx),
+        : [_] "{eax}" (leaf_id),
+          [_] "{ecx}" (subid),
     );
-
-    return cpuid_leaf;
+    return .{ .eax = eax, .ebx = ebx, .ecx = ecx, .edx = edx };
 }
 
 // Read control register 0 (XCR0). Used to detect features such as AVX.
 fn getXCR0() u32 {
     return asm volatile (
         \\ xor %%ecx, %%ecx
         \\ xgetbv
-        : [ret] "={eax}" (-> u32),
+        : [_] "={eax}" (-> u32),
         :
-        : "eax", "edx", "ecx"
+        : "edx", "ecx"
     );
 }