kroma-network · chokobole · May 16, 2024 · May 7, 2024 · May 7, 2024 · May 7, 2024
diff --git a/docs/how_to_use/how_to_build.md b/docs/how_to_use/how_to_build.md
@@ -232,6 +232,28 @@ export CXX=/usr/bin/g++-10
 export GCC_HOST_COMPILER_PATH=/usr/bin/gcc-10
 ```
 
+### Building with Clang on Ubuntu 22.04
+
+To utilize features like [always_inline](https://releases.llvm.org/15.0.0/tools/clang/docs/AttributeReference.html#always-inline-force-inline),
+you'll need clang++ at version 15 or higher. By default, Ubuntu 22.04 installs `clang-14` via `apt install clang`,
+so you'll need to manually install a newer version of clang and update `openmp` accordingly.
+
+First, install `clang-15` and the appropriate OpenMP libraries:
+
+```shell
+sudo apt install clang-15 libomp5-15 libomp-15-dev
+```
+
+Next, update `CC` and `CXX` to point to the newly installed `clang-15` and `clang++-15`:
+
+```shell
+export CC=/usr/bin/clang-15
+export CXX=/usr/bin/clang++-15
+```
+
+Make sure `CC` and `CXX` are properly updated to `clang-15` and `clang++-15`.
+These commands ensure that your build environment uses clang version 15 for compilation.
+
 ### Build CUDA with rust toolchain
 
 You may run into the following problem:

diff --git a/tachyon/base/openmp_util.h b/tachyon/base/openmp_util.h
@@ -9,11 +9,13 @@
 #endif  // defined(TACHYON_HAS_OPENMP)
 
 #if defined(TACHYON_HAS_OPENMP)
+#define OPENMP_CONSTEXPR
 #define OPENMP_PARALLEL_FOR(expr) _Pragma("omp parallel for") for (expr)
 #define OPENMP_PARALLEL_NESTED_FOR(expr) \
   _Pragma("omp parallel for collapse(2)") for (expr)
 #define OPENMP_FOR(expr) _Pragma("omp for") for (expr)
 #else
+#define OPENMP_CONSTEXPR constexpr
 #define OPENMP_PARALLEL_FOR(expr) for (expr)
 #define OPENMP_PARALLEL_NESTED_FOR(expr) for (expr)
 #define OPENMP_FOR(expr) for (expr)

diff --git a/tachyon/base/threading/platform_thread_linux.cc b/tachyon/base/threading/platform_thread_linux.cc
@@ -134,11 +134,11 @@ long sched_setattr(pid_t pid,
 #endif  // BUILDFLAG(IS_CHROMEOS)
 
 #if !BUILDFLAG(IS_NACL)
+/*
+TODO(chokobole):
 const char kCgroupDirectory[] =
     "/sys/fs/cgroup";
 
-/*
-TODO(chokobole):
 FilePath ThreadTypeToCgroupDirectory(const FilePath& cgroup_filepath,
                                      ThreadType thread_type) {
   switch (thread_type) {

diff --git a/tachyon/math/base/groups.h b/tachyon/math/base/groups.h
@@ -66,9 +66,9 @@ class MultiplicativeGroup : public MultiplicativeSemigroup<G> {
   // https://github.com/arkworks-rs/algebra/blob/5dfeedf560da6937a5de0a2163b7958bd32cd551/ff/src/fields/mod.rs#L355-L418.
   // Batch inverse: [a₁, a₂, ..., aₙ] -> [a₁⁻¹, a₂⁻¹, ... , aₙ⁻¹]
   template <typename InputContainer, typename OutputContainer>
-  [[nodiscard]] constexpr static bool BatchInverse(const InputContainer& groups,
-                                                   OutputContainer* inverses,
-                                                   const G& coeff = G::One()) {
+  [[nodiscard]] OPENMP_CONSTEXPR static bool BatchInverse(
+      const InputContainer& groups, OutputContainer* inverses,
+      const G& coeff = G::One()) {
     size_t size = std::size(groups);
     if (size != std::size(*inverses)) {
       LOG(ERROR) << "Size of |groups| and |inverses| do not match";

diff --git a/tachyon/math/base/semigroups.h b/tachyon/math/base/semigroups.h
@@ -437,9 +437,9 @@ class AdditiveSemigroup {
   // Multi Scalar Multi Base
   template <typename ScalarContainer, typename BaseContainer,
             typename OutputContainer>
-  constexpr static bool MultiScalarMulMSMB(const ScalarContainer& scalars,
-                                           const BaseContainer& bases,
-                                           OutputContainer* outputs) {
+  OPENMP_CONSTEXPR static bool MultiScalarMulMSMB(
+      const ScalarContainer& scalars, const BaseContainer& bases,
+      OutputContainer* outputs) {
     size_t size = scalars.size();
     if (size != std::size(bases)) return false;
     if (size != std::size(*outputs)) return false;
@@ -455,9 +455,8 @@ class AdditiveSemigroup {
 
   // Multi Scalar Single Base
   template <typename ScalarContainer, typename OutputContainer>
-  constexpr static bool MultiScalarMulMSSB(const ScalarContainer& scalars,
-                                           const G& base,
-                                           OutputContainer* outputs) {
+  OPENMP_CONSTEXPR static bool MultiScalarMulMSSB(
+      const ScalarContainer& scalars, const G& base, OutputContainer* outputs) {
     size_t size = std::size(scalars);
     if (size != std::size(*outputs)) return false;
     if (size == 0) {
@@ -472,9 +471,9 @@ class AdditiveSemigroup {
 
   // Single Scalar Multi Base
   template <typename Scalar, typename BaseContainer, typename OutputContainer>
-  constexpr static bool MultiScalarMulSSMB(const Scalar& scalar,
-                                           const BaseContainer& bases,
-                                           OutputContainer* outputs) {
+  OPENMP_CONSTEXPR static bool MultiScalarMulSSMB(const Scalar& scalar,
+                                                  const BaseContainer& bases,
+                                                  OutputContainer* outputs) {
     size_t size = std::size(bases);
     if (size != std::size(*outputs)) return false;
     if (size == 0) {

diff --git a/tachyon/math/elliptic_curves/msm/fixed_base_msm.h b/tachyon/math/elliptic_curves/msm/fixed_base_msm.h
@@ -144,7 +144,7 @@ class FixedBaseMSM {
                      std::move(outputs_first), std::move(outputs_last));
   }
 
-  constexpr void UpdateWindowTable(const Point& base) {
+  OPENMP_CONSTEXPR void UpdateWindowTable(const Point& base) {
     AddResult window_base;
     if constexpr (std::is_same_v<AddResult, Point>) {
       window_base = base;

diff --git a/tachyon/math/elliptic_curves/short_weierstrass/jacobian_point.h b/tachyon/math/elliptic_curves/short_weierstrass/jacobian_point.h
@@ -107,7 +107,7 @@ class JacobianPoint<
   }
 
   template <typename JacobianContainer, typename AffineContainer>
-  [[nodiscard]] constexpr static bool BatchNormalize(
+  [[nodiscard]] OPENMP_CONSTEXPR static bool BatchNormalize(
       const JacobianContainer& jacobian_points,
       AffineContainer* affine_points) {
     size_t size = std::size(jacobian_points);

diff --git a/tachyon/math/elliptic_curves/short_weierstrass/point_xyzz.h b/tachyon/math/elliptic_curves/short_weierstrass/point_xyzz.h
@@ -111,7 +111,7 @@ class PointXYZZ<_Curve,
   }
 
   template <typename PointXYZZContainer, typename AffineContainer>
-  [[nodiscard]] constexpr static bool BatchNormalize(
+  [[nodiscard]] OPENMP_CONSTEXPR static bool BatchNormalize(
       const PointXYZZContainer& point_xyzzs, AffineContainer* affine_points) {
     size_t size = std::size(point_xyzzs);
     if (size != std::size(*affine_points)) {
@@ -300,8 +300,8 @@ class PointXYZZ<_Curve,
  private:
   constexpr static void DoAdd(const PointXYZZ& a, const PointXYZZ& b,
                               PointXYZZ& c);
-  constexpr static void DoAdd(const PointXYZZ& a, const AffinePoint<Curve>& b,
-                              PointXYZZ& c);
+  OPENMP_CONSTEXPR static void DoAdd(const PointXYZZ& a,
+                                     const AffinePoint<Curve>& b, PointXYZZ& c);
   constexpr static void DoDoubleImpl(const PointXYZZ& a, PointXYZZ& b);
 
   BaseField x_;

diff --git a/tachyon/math/elliptic_curves/short_weierstrass/point_xyzz_impl.h b/tachyon/math/elliptic_curves/short_weierstrass/point_xyzz_impl.h
@@ -127,8 +127,8 @@ constexpr CLASS& CLASS::AddInPlace(const AffinePoint<Curve>& other) {
 
 // static
 template <typename Curve>
-constexpr void CLASS::DoAdd(const PointXYZZ& a, const AffinePoint<Curve>& b,
-                            PointXYZZ& c) {
+OPENMP_CONSTEXPR void CLASS::DoAdd(const PointXYZZ& a,
+                                   const AffinePoint<Curve>& b, PointXYZZ& c) {
   // https://hyperelliptic.org/EFD/g1p/auto-shortw-xyzz.html#addition-madd-2008-s
 
   // P = X2 * ZZ1 - X1

diff --git a/tachyon/math/elliptic_curves/short_weierstrass/projective_point.h b/tachyon/math/elliptic_curves/short_weierstrass/projective_point.h
@@ -105,7 +105,7 @@ class ProjectivePoint<
   }
 
   template <typename ProjectiveContainer, typename AffineContainer>
-  [[nodiscard]] constexpr static bool BatchNormalize(
+  [[nodiscard]] OPENMP_CONSTEXPR static bool BatchNormalize(
       const ProjectiveContainer& projective_points,
       AffineContainer* affine_points) {
     size_t size = std::size(projective_points);

diff --git a/tachyon/math/polynomials/univariate/mixed_radix_evaluation_domain.h b/tachyon/math/polynomials/univariate/mixed_radix_evaluation_domain.h
@@ -94,7 +94,7 @@ class MixedRadixEvaluationDomain
   }
 
   // UnivariateEvaluationDomain methods
-  constexpr void DoIFFT(DensePoly& poly) const override {
+  OPENMP_CONSTEXPR void DoIFFT(DensePoly& poly) const override {
     poly.coefficients_.coefficients_.resize(this->size_, F::Zero());
     BestFFT(poly, this->group_gen_inv_);
     if (this->offset_.IsOne()) {

diff --git a/tachyon/math/polynomials/univariate/radix2_evaluation_domain.h b/tachyon/math/polynomials/univariate/radix2_evaluation_domain.h
@@ -146,7 +146,7 @@ class Radix2EvaluationDomain : public UnivariateEvaluationDomain<F, MaxDegree> {
     FFTHelperInPlace(evals);
   }
 
-  constexpr void InOrderIFFTInPlace(DensePoly& poly) const {
+  OPENMP_CONSTEXPR void InOrderIFFTInPlace(DensePoly& poly) const {
     IFFTHelperInPlace(poly);
     if (this->offset_.IsOne()) {
       // clang-format off
@@ -179,8 +179,9 @@ class Radix2EvaluationDomain : public UnivariateEvaluationDomain<F, MaxDegree> {
   }
 
   template <FFTOrder Order, typename PolyOrEvals>
-  constexpr static void ApplyButterfly(PolyOrEvals& poly_or_evals,
-                                       absl::Span<const F> roots, size_t gap) {
+  OPENMP_CONSTEXPR static void ApplyButterfly(PolyOrEvals& poly_or_evals,
+                                              absl::Span<const F> roots,
+                                              size_t gap) {
     void (*fn)(F&, F&, const F&);
 
     if constexpr (Order == FFTOrder::kInOut) {
@@ -218,7 +219,7 @@ class Radix2EvaluationDomain : public UnivariateEvaluationDomain<F, MaxDegree> {
   //   [1],
   // ]
   // clang-format on
-  constexpr void PrepareRootsVecCache() {
+  OPENMP_CONSTEXPR void PrepareRootsVecCache() {
     if (this->log_size_of_group_ == 0) return;
 
     roots_vec_.resize(this->log_size_of_group_);

diff --git a/tachyon/math/polynomials/univariate/univariate_dense_coefficients.h b/tachyon/math/polynomials/univariate/univariate_dense_coefficients.h
@@ -170,7 +170,7 @@ class UnivariateDenseCoefficients {
   // multiplying either set of coefficients by a specified random field |r|,
   // and summing them together.
   template <bool MulRandomWithEvens>
-  constexpr UnivariateDenseCoefficients Fold(const Field& r) const {
+  OPENMP_CONSTEXPR UnivariateDenseCoefficients Fold(const Field& r) const {
     size_t size = coefficients_.size();
     std::vector<F> coefficients((size + 1) >> 1);
     OPENMP_PARALLEL_FOR(size_t i = 0; i < size; i += 2) {

diff --git a/tachyon/math/polynomials/univariate/univariate_evaluation_domain.h b/tachyon/math/polynomials/univariate/univariate_evaluation_domain.h
@@ -365,7 +365,7 @@ class UnivariateEvaluationDomain : public EvaluationDomain<F, MaxDegree> {
  protected:
   // Multiply the i-th element of |poly_or_evals| with |c|*|g|ⁱ.
   template <typename PolyOrEvals>
-  constexpr static void DistributePowersAndMulByConst(
+  OPENMP_CONSTEXPR static void DistributePowersAndMulByConst(
       PolyOrEvals& poly_or_evals, const F& g, const F& c) {
 #if defined(TACHYON_HAS_OPENMP)
     size_t thread_nums = static_cast<size_t>(omp_get_max_threads());
@@ -462,8 +462,8 @@ class UnivariateEvaluationDomain : public EvaluationDomain<F, MaxDegree> {
   }
 
   template <typename PolyOrEvals>
-  constexpr static void SwapElements(PolyOrEvals& poly_or_evals, size_t size,
-                                     uint32_t log_len) {
+  OPENMP_CONSTEXPR static void SwapElements(PolyOrEvals& poly_or_evals,
+                                            size_t size, uint32_t log_len) {
     OPENMP_PARALLEL_FOR(size_t idx = 1; idx < size; ++idx) {
       size_t ridx = base::bits::BitRev(idx) >> (sizeof(size_t) * 8 - log_len);
       if (idx < ridx) {

diff --git a/vendors/halo2/BUILD.bazel b/vendors/halo2/BUILD.bazel
@@ -1,8 +1,8 @@
 load("@crate_index//:defs.bzl", "aliases", "all_crate_deps")
 load("@cxx.rs//tools/bazel:rust_cxx_bridge.bzl", "rust_cxx_bridge")
 load("@local_config_cuda//cuda:build_defs.bzl", "if_cuda")
-load("//bazel:tachyon.bzl", "if_gpu_is_configured", "if_has_openmp")
-load("//bazel:tachyon_cc.bzl", "tachyon_cc_library")
+load("//bazel:tachyon.bzl", "if_gpu_is_configured")
+load("//bazel:tachyon_cc.bzl", "tachyon_cc_library", "tachyon_openmp_linkopts")
 load("//bazel:tachyon_rust.bzl", "tachyon_rust_library", "tachyon_rust_test")
 
 FEATURES = if_gpu_is_configured(["gpu"])
@@ -37,7 +37,7 @@ tachyon_rust_library(
 # rustc_flags = if_has_openmp(["-lgomp"]),
 cc_library(
     name = "openmp",
-    linkopts = if_has_openmp(["-fopenmp"]),
+    linkopts = tachyon_openmp_linkopts(),
 )
 
 tachyon_rust_test(