From b044efaa5123cc1498def6566b785a76184dfd99 Mon Sep 17 00:00:00 2001
From: Santeri Hannula <santeri@urbit.org>
Date: Wed, 9 Oct 2024 15:39:14 +0300
Subject: [PATCH] build: add build opts for enabling sanitizers

---
 INSTALL.md        | 22 +++++++++++++
 build.zig         | 81 ++++++++++++++++++++++++++++++++++++++++++++++-
 pkg/c3/portable.h |  6 +++-
 3 files changed, 107 insertions(+), 2 deletions(-)

diff --git a/INSTALL.md b/INSTALL.md
index 4e9ef81007..229154dc2e 100644
--- a/INSTALL.md
+++ b/INSTALL.md
@@ -28,6 +28,22 @@ pro hand -p true -s false -n false SIGBUS
 pro hand -p true -s false -n false SIGSEGV
 ```
 
+### Sanitizers
+
+Musl does not have proper sanitizer support, so until we support native (GNU) linux builds, these are limited to mac only.
+
+In order to use address sanitizer (`-Dasan`) and undefined behavior sanitizer (`-Dubsan`), you need to have version 18 of llvm installed on your machine. You also have to build natively, since cross-compilation is not supported.
+
+macOS:
+```terminal
+brew install llvm@18
+```
+
+<!-- linux: -->
+<!-- ```terminal -->
+<!-- apt-get install llvm-18 clang-18 -->
+<!-- ``` -->
+
 ## Build
 
 Once you install `zig`, you're ready to build:
@@ -73,6 +89,12 @@ Supported values:
 Provide additional compiler flags. These propagate to all build artifacts and
 dependencies.
 
+#### `-Dasan`
+Enable address sanitizer. Only supported in native macos builds. Requires llvm 18, see [prerequisites](#sanitizers).
+
+#### `-Dubsan`
+Enable undefined behavior sanitizer. Only supported in native macos builds. Requires llvm 18, see [prerequisites](#sanitizers).
+
 <!-- ## LSP Integration -->
 
 <!-- ```console -->
diff --git a/build.zig b/build.zig
index e6f1b25970..cc7458d880 100644
--- a/build.zig
+++ b/build.zig
@@ -2,6 +2,7 @@ const std = @import("std");
 
 const VERSION = "3.1";
 
+// TODO: x86_64-linux-gnu & aarch64-linux-gnu
 const targets: []const std.Target.Query = &.{
     .{ .cpu_arch = .aarch64, .os_tag = .macos, .abi = null },
     .{ .cpu_arch = .x86_64, .os_tag = .macos, .abi = null },
@@ -19,6 +20,8 @@ const BuildCfg = struct {
     mem_dbg: bool = false,
     c3dbg: bool = false,
     snapshot_validation: bool = false,
+    ubsan: bool = false,
+    asan: bool = false,
 };
 
 pub fn build(b: *std.Build) !void {
@@ -85,6 +88,26 @@ pub fn build(b: *std.Build) !void {
         "Binary name (Default: urbit)",
     ) orelse "urbit";
 
+    // Sanitizers are not properly supported in musl => mac only
+
+    const asan = if (target.query.isNative() and target.result.isDarwin())
+        b.option(
+            bool,
+            "asan",
+            "Enable address sanitizer (native only, requires llvm@18)",
+        ) orelse false
+    else
+        false;
+
+    const ubsan = if (target.query.isNative() and target.result.isDarwin())
+        b.option(
+            bool,
+            "ubsan",
+            "Enable undefined behavior sanitizer (native only, requires llvm@18)",
+        ) orelse false
+    else
+        false;
+
     // Parse short git rev
     //
     var file = try std.fs.cwd().openFile(".git/logs/HEAD", .{});
@@ -118,6 +141,8 @@ pub fn build(b: *std.Build) !void {
         .mem_dbg = mem_dbg,
         .c3dbg = c3dbg,
         .snapshot_validation = snapshot_validation,
+        .asan = asan,
+        .ubsan = ubsan,
         .include_test_steps = !all,
     };
 
@@ -156,12 +181,38 @@ fn build_single(
 
     try global_flags.appendSlice(cfg.flags);
     try global_flags.appendSlice(&.{
-        "-fno-sanitize=all",
         "-g",
         "-Wall",
         "-Werror",
     });
 
+    if (!cfg.asan and !cfg.ubsan)
+        try global_flags.appendSlice(&.{
+            "-fno-sanitize=all",
+        });
+
+    if (cfg.asan and !cfg.ubsan)
+        try global_flags.appendSlice(&.{
+            "-Wno-deprecated",
+            "-fsanitize=address",
+            "-fno-sanitize=undefined",
+            "-fno-sanitize-trap=undefined",
+        });
+
+    if (!cfg.asan and cfg.ubsan)
+        try global_flags.appendSlice(&.{
+            "-fsanitize=undefined",
+            "-fno-sanitize-trap=undefined",
+        });
+
+    if (cfg.asan and cfg.ubsan)
+        try global_flags.appendSlice(&.{
+            "-Wno-deprecated",
+            "-fsanitize=address",
+            "-fsanitize=undefined",
+            "-fno-sanitize-trap=undefined",
+        });
+
     //
     // CFLAGS for Urbit Libs and Binaries
     //
@@ -861,6 +912,34 @@ fn build_single(
 
     urbit.linkLibC();
 
+    if (t.isDarwin()) {
+        // Requires llvm@18 homebrew installation
+        if (cfg.asan or cfg.ubsan)
+            urbit.addLibraryPath(.{
+                .cwd_relative = "/opt/homebrew/opt/llvm/lib/clang/18/lib/darwin",
+            });
+        if (cfg.asan) urbit.linkSystemLibrary("clang_rt.asan_osx_dynamic");
+        if (cfg.ubsan) urbit.linkSystemLibrary("clang_rt.ubsan_osx_dynamic");
+    }
+
+    if (t.os.tag == .linux) {
+        // Requires llvm-18 and clang-18 installation
+        if (cfg.asan or cfg.ubsan)
+            urbit.addLibraryPath(.{
+                .cwd_relative = "/usr/lib/clang/18/lib/linux",
+            });
+        if (t.cpu.arch == .x86_64) {
+            if (cfg.asan) urbit.linkSystemLibrary("clang_rt.asan-x86_64");
+            if (cfg.ubsan)
+                urbit.linkSystemLibrary("clang_rt.ubsan_standalone-x86_64");
+        }
+        if (t.cpu.arch == .aarch64) {
+            if (cfg.asan) urbit.linkSystemLibrary("clang_rt.asan-aarch64");
+            if (cfg.ubsan)
+                urbit.linkSystemLibrary("clang_rt.ubsan_standalone-aarch64");
+        }
+    }
+
     urbit.linkLibrary(vere);
     urbit.linkLibrary(pkg_noun);
     urbit.linkLibrary(pkg_c3);
diff --git a/pkg/c3/portable.h b/pkg/c3/portable.h
index 4a507a8e3e..eeee55cc22 100644
--- a/pkg/c3/portable.h
+++ b/pkg/c3/portable.h
@@ -129,7 +129,11 @@
 #       define U3_OS_LoomBits 30
 #   elif defined(U3_OS_osx)
 #     ifdef __LP64__
-#       define U3_OS_LoomBase 0x28000000000
+#       ifdef ASAN_ENABLED
+#         define U3_OS_LoomBase 0x728000000000
+#       else
+#         define U3_OS_LoomBase 0x28000000000
+#       endif
 #     else
 #       define U3_OS_LoomBase 0x4000000
 #     endif