diff --git a/clang/test/CodeGen/PowerPC/aix-vaargs.c b/clang/test/CodeGen/PowerPC/aix-vaargs.c
@@ -35,7 +35,7 @@ void testva (int n, ...) {
 
 // CHECK-NEXT:  %v = alloca i32, align 4
 // CHECK-NEXT:  store i32 %n, ptr %n.addr, align 4
-// CHECK-NEXT:  call void @llvm.va_start(ptr %ap)
+// CHECK-NEXT:  call void @llvm.va_start.p0(ptr %ap)
 
 // AIX32-NEXT:  %argp.cur = load ptr, ptr %ap, align 4
 // AIX32-NEXT:  %argp.next = getelementptr inbounds i8, ptr %argp.cur, i32 16
@@ -48,7 +48,7 @@ void testva (int n, ...) {
 // AIX32-NEXT:  call void @llvm.memcpy.p0.p0.i32(ptr align 8 %t, ptr align 4 %argp.cur, i32 16, i1 false)
 // AIX64-NEXT:  call void @llvm.memcpy.p0.p0.i64(ptr align 8 %t, ptr align 8 %argp.cur, i64 16, i1 false)
 
-// CHECK-NEXT:  call void @llvm.va_copy(ptr %ap2, ptr %ap)
+// CHECK-NEXT:  call void @llvm.va_copy.p0(ptr %ap2, ptr %ap)
 
 // AIX32-NEXT:  %argp.cur1 = load ptr, ptr %ap2, align 4
 // AIX32-NEXT:  %argp.next2 = getelementptr inbounds i8, ptr %argp.cur1, i32 4
@@ -62,14 +62,14 @@ void testva (int n, ...) {
 // AIX64-NEXT:  %1 = load i32, ptr %0, align 4
 // AIX64-NEXT:  store i32 %1, ptr %v, align 4
 
-// CHECK-NEXT:  call void @llvm.va_end(ptr %ap2)
-// CHECK-NEXT:  call void @llvm.va_end(ptr %ap)
+// CHECK-NEXT:  call void @llvm.va_end.p0(ptr %ap2)
+// CHECK-NEXT:  call void @llvm.va_end.p0(ptr %ap)
 // CHECK-NEXT:  ret void
 
-// CHECK: declare void @llvm.va_start(ptr)
+// CHECK: declare void @llvm.va_start.p0(ptr)
 
 // AIX32: declare void @llvm.memcpy.p0.p0.i32(ptr noalias nocapture writeonly, ptr noalias nocapture readonly, i32, i1 immarg)
 // AIX64: declare void @llvm.memcpy.p0.p0.i64(ptr noalias nocapture writeonly, ptr noalias nocapture readonly, i64, i1 immarg)
 
-// CHECK: declare void @llvm.va_copy(ptr, ptr)
-// CHECK: declare void @llvm.va_end(ptr)
+// CHECK: declare void @llvm.va_copy.p0(ptr, ptr)
+// CHECK: declare void @llvm.va_end.p0(ptr)
diff --git a/clang/test/CodeGen/PowerPC/ppc64le-varargs-f128.c b/clang/test/CodeGen/PowerPC/ppc64le-varargs-f128.c
@@ -31,7 +31,7 @@ void foo_ls(ldbl128_s);
 // OMP-TARGET: call void @foo_ld(ppc_fp128 noundef %[[V3]])
 
 // OMP-HOST-LABEL: define{{.*}} void @omp(
-// OMP-HOST: call void @llvm.va_start(ptr %[[AP:[0-9a-zA-Z_.]+]])
+// OMP-HOST: call void @llvm.va_start.p0(ptr %[[AP:[0-9a-zA-Z_.]+]])
 // OMP-HOST: %[[CUR:[0-9a-zA-Z_.]+]] = load ptr, ptr %[[AP]], align 8
 // OMP-HOST: %[[TMP0:[^ ]+]] = getelementptr inbounds i8, ptr %[[CUR]], i32 15
 // OMP-HOST: %[[ALIGN:[^ ]+]] = call ptr @llvm.ptrmask.p0.i64(ptr %[[TMP0]], i64 -16)
@@ -49,13 +49,13 @@ void omp(int n, ...) {
 }
 
 // IEEE-LABEL: define{{.*}} void @f128
-// IEEE: call void @llvm.va_start(ptr %[[AP:[0-9a-zA-Z_.]+]])
+// IEEE: call void @llvm.va_start.p0(ptr %[[AP:[0-9a-zA-Z_.]+]])
 // IEEE: %[[CUR:[0-9a-zA-Z_.]+]] = load ptr, ptr %[[AP]]
 // IEEE: %[[TMP0:[^ ]+]] = getelementptr inbounds i8, ptr %[[CUR]], i32 15
 // IEEE: %[[ALIGN:[^ ]+]] = call ptr @llvm.ptrmask.p0.i64(ptr %[[TMP0]], i64 -16)
 // IEEE: %[[V4:[0-9a-zA-Z_.]+]] = load fp128, ptr %[[ALIGN]], align 16
 // IEEE: call void @foo_fq(fp128 noundef %[[V4]])
-// IEEE: call void @llvm.va_end(ptr %[[AP]])
+// IEEE: call void @llvm.va_end.p0(ptr %[[AP]])
 void f128(int n, ...) {
   va_list ap;
   va_start(ap, n);
@@ -64,20 +64,20 @@ void f128(int n, ...) {
 }
 
 // IEEE-LABEL: define{{.*}} void @long_double
-// IEEE: call void @llvm.va_start(ptr %[[AP:[0-9a-zA-Z_.]+]])
+// IEEE: call void @llvm.va_start.p0(ptr %[[AP:[0-9a-zA-Z_.]+]])
 // IEEE: %[[CUR:[0-9a-zA-Z_.]+]] = load ptr, ptr %[[AP]]
 // IEEE: %[[TMP0:[^ ]+]] = getelementptr inbounds i8, ptr %[[CUR]], i32 15
 // IEEE: %[[ALIGN:[^ ]+]] = call ptr @llvm.ptrmask.p0.i64(ptr %[[TMP0]], i64 -16)
 // IEEE: %[[V4:[0-9a-zA-Z_.]+]] = load fp128, ptr %[[ALIGN]], align 16
 // IEEE: call void @foo_ld(fp128 noundef %[[V4]])
-// IEEE: call void @llvm.va_end(ptr %[[AP]])
+// IEEE: call void @llvm.va_end.p0(ptr %[[AP]])
 
 // IBM-LABEL: define{{.*}} void @long_double
-// IBM: call void @llvm.va_start(ptr  %[[AP:[0-9a-zA-Z_.]+]])
+// IBM: call void @llvm.va_start.p0(ptr  %[[AP:[0-9a-zA-Z_.]+]])
 // IBM: %[[CUR:[0-9a-zA-Z_.]+]] = load ptr, ptr %[[AP]]
 // IBM: %[[V4:[0-9a-zA-Z_.]+]] = load ppc_fp128, ptr %[[CUR]], align 8
 // IBM: call void @foo_ld(ppc_fp128 noundef %[[V4]])
-// IBM: call void @llvm.va_end(ptr %[[AP]])
+// IBM: call void @llvm.va_end.p0(ptr %[[AP]])
 void long_double(int n, ...) {
   va_list ap;
   va_start(ap, n);
@@ -86,7 +86,7 @@ void long_double(int n, ...) {
 }
 
 // IEEE-LABEL: define{{.*}} void @long_double_struct
-// IEEE: call void @llvm.va_start(ptr %[[AP:[0-9a-zA-Z_.]+]])
+// IEEE: call void @llvm.va_start.p0(ptr %[[AP:[0-9a-zA-Z_.]+]])
 // IEEE: %[[CUR:[0-9a-zA-Z_.]+]] = load ptr, ptr %[[AP]]
 // IEEE: %[[TMP0:[^ ]+]] = getelementptr inbounds i8, ptr %[[CUR]], i32 15
 // IEEE: %[[ALIGN:[^ ]+]] = call ptr @llvm.ptrmask.p0.i64(ptr %[[TMP0]], i64 -16)
@@ -96,7 +96,7 @@ void long_double(int n, ...) {
 // IEEE: %[[COERCE:[0-9a-zA-Z_.]+]] = getelementptr inbounds %struct.ldbl128_s, ptr %[[TMP]], i32 0, i32 0
 // IEEE: %[[V4:[0-9a-zA-Z_.]+]] = load fp128, ptr %[[COERCE]], align 16
 // IEEE: call void @foo_ls(fp128 inreg %[[V4]])
-// IEEE: call void @llvm.va_end(ptr %[[AP]])
+// IEEE: call void @llvm.va_end.p0(ptr %[[AP]])
 void long_double_struct(int n, ...) {
   va_list ap;
   va_start(ap, n);

diff --git a/clang/test/CodeGen/RISCV/riscv-func-attr-target-err.c b/clang/test/CodeGen/RISCV/riscv-func-attr-target-err.c
@@ -2,6 +2,28 @@
 // RUN: not %clang_cc1 -triple riscv64 -target-feature +zifencei -target-feature +m -target-feature +a \
 // RUN:  -emit-llvm %s 2>&1 | FileCheck %s
 
+#include <riscv_vector.h>
+
+void test_builtin() {
+// CHECK: error: '__builtin_rvv_vsetvli' needs target feature zve32x
+  __riscv_vsetvl_e8m8(1);
+}
+
+void test_rvv_i32_type() {
+// CHECK: error: RISC-V type 'vint32m1_t' (aka '__rvv_int32m1_t') requires the 'zve32x' extension
+  vint32m1_t v;
+}
+
+void test_rvv_f32_type() {
+// CHECK: error: RISC-V type 'vfloat32m1_t' (aka '__rvv_float32m1_t') requires the 'zve32f' extension
+  vfloat32m1_t v;
+}
+
+void test_rvv_f64_type() {
+// CHECK: error: RISC-V type 'vfloat64m1_t' (aka '__rvv_float64m1_t') requires the 'zve64d' extension
+  vfloat64m1_t v;
+}
+
 // CHECK: error: duplicate 'arch=' in the 'target' attribute string;
 __attribute__((target("arch=rv64gc;arch=rv64gc_zbb"))) void testMultiArchSelectLast() {}
 // CHECK: error: duplicate 'cpu=' in the 'target' attribute string;

diff --git a/clang/test/CodeGen/RISCV/riscv-func-attr-target.c b/clang/test/CodeGen/RISCV/riscv-func-attr-target.c
@@ -4,6 +4,8 @@
 // RUN:  -target-feature -relax -target-feature -zfa \
 // RUN:  -emit-llvm %s -o - | FileCheck %s
 
+#include <riscv_vector.h>
+
 // CHECK-LABEL: define dso_local void @testDefault
 // CHECK-SAME: () #0 {
 void testDefault() {}
@@ -35,6 +37,34 @@ testAttrFullArchAndAttrCpu() {}
 // CHECK-SAME: () #8 {
 __attribute__((target("cpu=sifive-u54"))) void testAttrCpuOnly() {}
 
+__attribute__((target("arch=+zve32x")))
+void test_builtin_w_zve32x() {
+// CHECK-LABEL: test_builtin_w_zve32x
+// CHECK-SAME: #9
+  __riscv_vsetvl_e8m8(1);
+}
+
+__attribute__((target("arch=+zve32x")))
+void test_rvv_i32_type_w_zve32x() {
+// CHECK-LABEL: test_rvv_i32_type_w_zve32x
+// CHECK-SAME: #9
+  vint32m1_t v;
+}
+
+__attribute__((target("arch=+zve32f")))
+void test_rvv_f32_type_w_zve32f() {
+// CHECK-LABEL: test_rvv_f32_type_w_zve32f
+// CHECK-SAME: #11
+  vfloat32m1_t v;
+}
+
+__attribute__((target("arch=+zve64d")))
+void test_rvv_f64_type_w_zve64d() {
+// CHECK-LABEL: test_rvv_f64_type_w_zve64d
+// CHECK-SAME: #12
+  vfloat64m1_t v;
+}
+
 //.
 // CHECK: attributes #0 = { {{.*}}"target-features"="+64bit,+a,+m,+save-restore,+zifencei,-relax,-zbb,-zfa" }
 // CHECK: attributes #1 = { {{.*}}"target-cpu"="rocket-rv64" "target-features"="+64bit,+a,+d,+f,+m,+save-restore,+v,+zicsr,+zifencei,+zve32f,+zve32x,+zve64d,+zve64f,+zve64x,+zvl128b,+zvl32b,+zvl64b,-relax,-zbb,-zfa" "tune-cpu"="generic-rv64" }
@@ -46,3 +76,6 @@ __attribute__((target("cpu=sifive-u54"))) void testAttrCpuOnly() {}
 // CHECK: attributes #6 = { {{.*}}"target-cpu"="sifive-u54" "target-features"="+64bit,+a,+m,+save-restore,+zbb,+zifencei,-relax,-zfa" }
 // CHECK: attributes #7 = { {{.*}}"target-cpu"="sifive-u54" "target-features"="+64bit,+m,+save-restore,{{(-[[:alnum:]-]+)(,-[[:alnum:]-]+)*}}" }
 // CHECK: attributes #8 = { {{.*}}"target-cpu"="sifive-u54" "target-features"="+64bit,+a,+c,+d,+f,+m,+save-restore,+zicsr,+zifencei,{{(-[[:alnum:]-]+)(,-[[:alnum:]-]+)*}}" }
+// CHECK: attributes #9 = { {{.*}}"target-features"="+64bit,+a,+m,+save-restore,+zicsr,+zifencei,+zve32x,+zvl32b,-relax,-zbb,-zfa" }
+// CHECK: attributes #11 = { {{.*}}"target-features"="+64bit,+a,+f,+m,+save-restore,+zicsr,+zifencei,+zve32f,+zve32x,+zvl32b,-relax,-zbb,-zfa" }
+// CHECK: attributes #12 = { {{.*}}"target-features"="+64bit,+a,+d,+f,+m,+save-restore,+zicsr,+zifencei,+zve32f,+zve32x,+zve64d,+zve64f,+zve64x,+zvl32b,+zvl64b,-relax,-zbb,-zfa" }
diff --git a/clang/test/CodeGen/RISCV/riscv-vector-callingconv-llvm-ir.c b/clang/test/CodeGen/RISCV/riscv-vector-callingconv-llvm-ir.c
@@ -0,0 +1,34 @@
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -emit-llvm %s -o - | FileCheck -check-prefix=CHECK-LLVM %s
+// RUN: %clang_cc1 -std=c23 -triple riscv64 -target-feature +v \
+// RUN:   -emit-llvm %s -o - | FileCheck -check-prefix=CHECK-LLVM %s
+
+#include <riscv_vector.h>
+
+// CHECK-LLVM: call riscv_vector_cc <vscale x 2 x i32> @bar
+vint32m1_t __attribute__((riscv_vector_cc)) bar(vint32m1_t input);
+vint32m1_t test_vector_cc_attr(vint32m1_t input, int32_t *base, size_t vl) {
+  vint32m1_t val = __riscv_vle32_v_i32m1(base, vl);
+  vint32m1_t ret = bar(input);
+  __riscv_vse32_v_i32m1(base, val, vl);
+  return ret;
+}
+
+// CHECK-LLVM: call riscv_vector_cc <vscale x 2 x i32> @bar
+[[riscv::vector_cc]] vint32m1_t bar(vint32m1_t input);
+vint32m1_t test_vector_cc_attr2(vint32m1_t input, int32_t *base, size_t vl) {
+  vint32m1_t val = __riscv_vle32_v_i32m1(base, vl);
+  vint32m1_t ret = bar(input);
+  __riscv_vse32_v_i32m1(base, val, vl);
+  return ret;
+}
+
+// CHECK-LLVM: call <vscale x 2 x i32> @baz
+vint32m1_t baz(vint32m1_t input);
+vint32m1_t test_no_vector_cc_attr(vint32m1_t input, int32_t *base, size_t vl) {
+  vint32m1_t val = __riscv_vle32_v_i32m1(base, vl);
+  vint32m1_t ret = baz(input);
+  __riscv_vse32_v_i32m1(base, val, vl);
+  return ret;
+}
diff --git a/clang/test/CodeGen/RISCV/riscv-vector-callingconv-llvm-ir.cpp b/clang/test/CodeGen/RISCV/riscv-vector-callingconv-llvm-ir.cpp
@@ -0,0 +1,32 @@
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -std=c++11 -triple riscv64 -target-feature +v \
+// RUN:   -emit-llvm %s -o - | FileCheck -check-prefix=CHECK-LLVM %s
+
+#include <riscv_vector.h>
+
+// CHECK-LLVM: call riscv_vector_cc <vscale x 2 x i32> @_Z3baru15__rvv_int32m1_t
+vint32m1_t __attribute__((riscv_vector_cc)) bar(vint32m1_t input);
+vint32m1_t test_vector_cc_attr(vint32m1_t input, int32_t *base, size_t vl) {
+  vint32m1_t val = __riscv_vle32_v_i32m1(base, vl);
+  vint32m1_t ret = bar(input);
+  __riscv_vse32_v_i32m1(base, val, vl);
+  return ret;
+}
+
+// CHECK-LLVM: call riscv_vector_cc <vscale x 2 x i32> @_Z3baru15__rvv_int32m1_t
+[[riscv::vector_cc]] vint32m1_t bar(vint32m1_t input);
+vint32m1_t test_vector_cc_attr2(vint32m1_t input, int32_t *base, size_t vl) {
+  vint32m1_t val = __riscv_vle32_v_i32m1(base, vl);
+  vint32m1_t ret = bar(input);
+  __riscv_vse32_v_i32m1(base, val, vl);
+  return ret;
+}
+
+// CHECK-LLVM: call <vscale x 2 x i32> @_Z3bazu15__rvv_int32m1_t
+vint32m1_t baz(vint32m1_t input);
+vint32m1_t test_no_vector_cc_attr(vint32m1_t input, int32_t *base, size_t vl) {
+  vint32m1_t val = __riscv_vle32_v_i32m1(base, vl);
+  vint32m1_t ret = baz(input);
+  __riscv_vse32_v_i32m1(base, val, vl);
+  return ret;
+}
diff --git a/clang/test/CodeGen/RISCV/riscv-vector-callingconv.c b/clang/test/CodeGen/RISCV/riscv-vector-callingconv.c
@@ -0,0 +1,17 @@
+// RUN: %clang_cc1 %s -std=c23 -triple riscv64 -target-feature +v -verify
+
+__attribute__((riscv_vector_cc)) int var; // expected-warning {{'riscv_vector_cc' only applies to function types; type here is 'int'}}
+
+__attribute__((riscv_vector_cc)) void func();
+__attribute__((riscv_vector_cc(1))) void func_invalid(); // expected-error {{'riscv_vector_cc' attribute takes no arguments}}
+
+void test_no_attribute(int); // expected-note {{previous declaration is here}}
+void __attribute__((riscv_vector_cc)) test_no_attribute(int x) { } // expected-error {{function declared 'riscv_vector_cc' here was previously declared without calling convention}}
+
+[[riscv::vector_cc]] int var2; // expected-warning {{'vector_cc' only applies to function types; type here is 'int'}}
+
+[[riscv::vector_cc]] void func2();
+[[riscv::vector_cc(1)]] void func_invalid2(); // expected-error {{'vector_cc' attribute takes no arguments}}
+
+void test_no_attribute2(int); // expected-note {{previous declaration is here}}
+[[riscv::vector_cc]] void test_no_attribute2(int x) { } // expected-error {{function declared 'riscv_vector_cc' here was previously declared without calling convention}}
diff --git a/clang/test/CodeGen/RISCV/riscv-vector-callingconv.cpp b/clang/test/CodeGen/RISCV/riscv-vector-callingconv.cpp
@@ -0,0 +1,35 @@
+// RUN: %clang_cc1 %s -triple riscv64 -target-feature +v -verify
+
+__attribute__((riscv_vector_cc)) int var; // expected-warning {{'riscv_vector_cc' only applies to function types; type here is 'int'}}
+
+__attribute__((riscv_vector_cc)) void func();
+__attribute__((riscv_vector_cc(1))) void func_invalid(); // expected-error {{'riscv_vector_cc' attribute takes no arguments}}
+
+void test_no_attribute(int); // expected-note {{previous declaration is here}}
+void __attribute__((riscv_vector_cc)) test_no_attribute(int x) { } // expected-error {{function declared 'riscv_vector_cc' here was previously declared without calling convention}}
+
+class test_cc {
+  __attribute__((riscv_vector_cc)) void member_func();
+};
+
+void test_lambda() {
+  __attribute__((riscv_vector_cc)) auto lambda = []() { // expected-warning {{'riscv_vector_cc' only applies to function types; type here is 'auto'}}
+  };
+}
+
+[[riscv::vector_cc]] int var2; // expected-warning {{'vector_cc' only applies to function types; type here is 'int'}}
+
+[[riscv::vector_cc]] void func2();
+[[riscv::vector_cc(1)]] void func_invalid2(); // expected-error {{'vector_cc' attribute takes no arguments}}
+
+void test_no_attribute2(int); // expected-note {{previous declaration is here}}
+[[riscv::vector_cc]] void test_no_attribute2(int x) { } // expected-error {{function declared 'riscv_vector_cc' here was previously declared without calling convention}}
+
+class test_cc2 {
+  [[riscv::vector_cc]] void member_func();
+};
+
+void test_lambda2() {
+  [[riscv::vector_cc]] auto lambda = []() { // expected-warning {{'vector_cc' only applies to function types; type here is 'auto'}}
+  };
+}
diff --git a/clang/test/CodeGen/RISCV/riscv32-vararg.c b/clang/test/CodeGen/RISCV/riscv32-vararg.c
@@ -80,13 +80,13 @@ void f_va_caller(void) {
 // CHECK-NEXT:    [[VA:%.*]] = alloca ptr, align 4
 // CHECK-NEXT:    [[V:%.*]] = alloca i32, align 4
 // CHECK-NEXT:    store ptr [[FMT]], ptr [[FMT_ADDR]], align 4
-// CHECK-NEXT:    call void @llvm.va_start(ptr [[VA]])
+// CHECK-NEXT:    call void @llvm.va_start.p0(ptr [[VA]])
 // CHECK-NEXT:    [[ARGP_CUR:%.*]] = load ptr, ptr [[VA]], align 4
 // CHECK-NEXT:    [[ARGP_NEXT:%.*]] = getelementptr inbounds i8, ptr [[ARGP_CUR]], i32 4
 // CHECK-NEXT:    store ptr [[ARGP_NEXT]], ptr [[VA]], align 4
 // CHECK-NEXT:    [[TMP0:%.*]] = load i32, ptr [[ARGP_CUR]], align 4
 // CHECK-NEXT:    store i32 [[TMP0]], ptr [[V]], align 4
-// CHECK-NEXT:    call void @llvm.va_end(ptr [[VA]])
+// CHECK-NEXT:    call void @llvm.va_end.p0(ptr [[VA]])
 // CHECK-NEXT:    [[TMP1:%.*]] = load i32, ptr [[V]], align 4
 // CHECK-NEXT:    ret i32 [[TMP1]]
 //
@@ -111,15 +111,15 @@ int f_va_1(char *fmt, ...) {
 // CHECK-ILP32F-NEXT:    [[VA:%.*]] = alloca ptr, align 4
 // CHECK-ILP32F-NEXT:    [[V:%.*]] = alloca double, align 8
 // CHECK-ILP32F-NEXT:    store ptr [[FMT]], ptr [[FMT_ADDR]], align 4
-// CHECK-ILP32F-NEXT:    call void @llvm.va_start(ptr [[VA]])
+// CHECK-ILP32F-NEXT:    call void @llvm.va_start.p0(ptr [[VA]])
 // CHECK-ILP32F-NEXT:    [[ARGP_CUR:%.*]] = load ptr, ptr [[VA]], align 4
 // CHECK-ILP32F-NEXT:    [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[ARGP_CUR]], i32 7
 // CHECK-ILP32F-NEXT:    [[ARGP_CUR_ALIGNED:%.*]] = call ptr @llvm.ptrmask.p0.i32(ptr [[TMP0]], i32 -8)
 // CHECK-ILP32F-NEXT:    [[ARGP_NEXT:%.*]] = getelementptr inbounds i8, ptr [[ARGP_CUR_ALIGNED]], i32 8
 // CHECK-ILP32F-NEXT:    store ptr [[ARGP_NEXT]], ptr [[VA]], align 4
 // CHECK-ILP32F-NEXT:    [[TMP1:%.*]] = load double, ptr [[ARGP_CUR_ALIGNED]], align 8
 // CHECK-ILP32F-NEXT:    store double [[TMP1]], ptr [[V]], align 8
-// CHECK-ILP32F-NEXT:    call void @llvm.va_end(ptr [[VA]])
+// CHECK-ILP32F-NEXT:    call void @llvm.va_end.p0(ptr [[VA]])
 // CHECK-ILP32F-NEXT:    [[TMP2:%.*]] = load double, ptr [[V]], align 8
 // CHECK-ILP32F-NEXT:    ret double [[TMP2]]
 //
@@ -130,15 +130,15 @@ int f_va_1(char *fmt, ...) {
 // CHECK-ILP32D-NEXT:    [[VA:%.*]] = alloca ptr, align 4
 // CHECK-ILP32D-NEXT:    [[V:%.*]] = alloca double, align 8
 // CHECK-ILP32D-NEXT:    store ptr [[FMT]], ptr [[FMT_ADDR]], align 4
-// CHECK-ILP32D-NEXT:    call void @llvm.va_start(ptr [[VA]])
+// CHECK-ILP32D-NEXT:    call void @llvm.va_start.p0(ptr [[VA]])
 // CHECK-ILP32D-NEXT:    [[ARGP_CUR:%.*]] = load ptr, ptr [[VA]], align 4
 // CHECK-ILP32D-NEXT:    [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[ARGP_CUR]], i32 7
 // CHECK-ILP32D-NEXT:    [[ARGP_CUR_ALIGNED:%.*]] = call ptr @llvm.ptrmask.p0.i32(ptr [[TMP0]], i32 -8)
 // CHECK-ILP32D-NEXT:    [[ARGP_NEXT:%.*]] = getelementptr inbounds i8, ptr [[ARGP_CUR_ALIGNED]], i32 8
 // CHECK-ILP32D-NEXT:    store ptr [[ARGP_NEXT]], ptr [[VA]], align 4
 // CHECK-ILP32D-NEXT:    [[TMP1:%.*]] = load double, ptr [[ARGP_CUR_ALIGNED]], align 8
 // CHECK-ILP32D-NEXT:    store double [[TMP1]], ptr [[V]], align 8
-// CHECK-ILP32D-NEXT:    call void @llvm.va_end(ptr [[VA]])
+// CHECK-ILP32D-NEXT:    call void @llvm.va_end.p0(ptr [[VA]])
 // CHECK-ILP32D-NEXT:    [[TMP2:%.*]] = load double, ptr [[V]], align 8
 // CHECK-ILP32D-NEXT:    ret double [[TMP2]]
 //
@@ -149,13 +149,13 @@ int f_va_1(char *fmt, ...) {
 // CHECK-ILP32E-NEXT:    [[VA:%.*]] = alloca ptr, align 4
 // CHECK-ILP32E-NEXT:    [[V:%.*]] = alloca double, align 8
 // CHECK-ILP32E-NEXT:    store ptr [[FMT]], ptr [[FMT_ADDR]], align 4
-// CHECK-ILP32E-NEXT:    call void @llvm.va_start(ptr [[VA]])
+// CHECK-ILP32E-NEXT:    call void @llvm.va_start.p0(ptr [[VA]])
 // CHECK-ILP32E-NEXT:    [[ARGP_CUR:%.*]] = load ptr, ptr [[VA]], align 4
 // CHECK-ILP32E-NEXT:    [[ARGP_NEXT:%.*]] = getelementptr inbounds i8, ptr [[ARGP_CUR]], i32 8
 // CHECK-ILP32E-NEXT:    store ptr [[ARGP_NEXT]], ptr [[VA]], align 4
 // CHECK-ILP32E-NEXT:    [[TMP0:%.*]] = load double, ptr [[ARGP_CUR]], align 4
 // CHECK-ILP32E-NEXT:    store double [[TMP0]], ptr [[V]], align 8
-// CHECK-ILP32E-NEXT:    call void @llvm.va_end(ptr [[VA]])
+// CHECK-ILP32E-NEXT:    call void @llvm.va_end.p0(ptr [[VA]])
 // CHECK-ILP32E-NEXT:    [[TMP1:%.*]] = load double, ptr [[V]], align 8
 // CHECK-ILP32E-NEXT:    ret double [[TMP1]]
 //
@@ -180,7 +180,7 @@ double f_va_2(char *fmt, ...) {
 // CHECK-ILP32F-NEXT:    [[W:%.*]] = alloca i32, align 4
 // CHECK-ILP32F-NEXT:    [[X:%.*]] = alloca double, align 8
 // CHECK-ILP32F-NEXT:    store ptr [[FMT]], ptr [[FMT_ADDR]], align 4
-// CHECK-ILP32F-NEXT:    call void @llvm.va_start(ptr [[VA]])
+// CHECK-ILP32F-NEXT:    call void @llvm.va_start.p0(ptr [[VA]])
 // CHECK-ILP32F-NEXT:    [[ARGP_CUR:%.*]] = load ptr, ptr [[VA]], align 4
 // CHECK-ILP32F-NEXT:    [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[ARGP_CUR]], i32 7
 // CHECK-ILP32F-NEXT:    [[ARGP_CUR_ALIGNED:%.*]] = call ptr @llvm.ptrmask.p0.i32(ptr [[TMP0]], i32 -8)
@@ -200,7 +200,7 @@ double f_va_2(char *fmt, ...) {
 // CHECK-ILP32F-NEXT:    store ptr [[ARGP_NEXT4]], ptr [[VA]], align 4
 // CHECK-ILP32F-NEXT:    [[TMP4:%.*]] = load double, ptr [[ARGP_CUR3_ALIGNED]], align 8
 // CHECK-ILP32F-NEXT:    store double [[TMP4]], ptr [[X]], align 8
-// CHECK-ILP32F-NEXT:    call void @llvm.va_end(ptr [[VA]])
+// CHECK-ILP32F-NEXT:    call void @llvm.va_end.p0(ptr [[VA]])
 // CHECK-ILP32F-NEXT:    [[TMP5:%.*]] = load double, ptr [[V]], align 8
 // CHECK-ILP32F-NEXT:    [[TMP6:%.*]] = load double, ptr [[X]], align 8
 // CHECK-ILP32F-NEXT:    [[ADD:%.*]] = fadd double [[TMP5]], [[TMP6]]
@@ -215,7 +215,7 @@ double f_va_2(char *fmt, ...) {
 // CHECK-ILP32D-NEXT:    [[W:%.*]] = alloca i32, align 4
 // CHECK-ILP32D-NEXT:    [[X:%.*]] = alloca double, align 8
 // CHECK-ILP32D-NEXT:    store ptr [[FMT]], ptr [[FMT_ADDR]], align 4
-// CHECK-ILP32D-NEXT:    call void @llvm.va_start(ptr [[VA]])
+// CHECK-ILP32D-NEXT:    call void @llvm.va_start.p0(ptr [[VA]])
 // CHECK-ILP32D-NEXT:    [[ARGP_CUR:%.*]] = load ptr, ptr [[VA]], align 4
 // CHECK-ILP32D-NEXT:    [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[ARGP_CUR]], i32 7
 // CHECK-ILP32D-NEXT:    [[ARGP_CUR_ALIGNED:%.*]] = call ptr @llvm.ptrmask.p0.i32(ptr [[TMP0]], i32 -8)
@@ -235,7 +235,7 @@ double f_va_2(char *fmt, ...) {
 // CHECK-ILP32D-NEXT:    store ptr [[ARGP_NEXT4]], ptr [[VA]], align 4
 // CHECK-ILP32D-NEXT:    [[TMP4:%.*]] = load double, ptr [[ARGP_CUR3_ALIGNED]], align 8
 // CHECK-ILP32D-NEXT:    store double [[TMP4]], ptr [[X]], align 8
-// CHECK-ILP32D-NEXT:    call void @llvm.va_end(ptr [[VA]])
+// CHECK-ILP32D-NEXT:    call void @llvm.va_end.p0(ptr [[VA]])
 // CHECK-ILP32D-NEXT:    [[TMP5:%.*]] = load double, ptr [[V]], align 8
 // CHECK-ILP32D-NEXT:    [[TMP6:%.*]] = load double, ptr [[X]], align 8
 // CHECK-ILP32D-NEXT:    [[ADD:%.*]] = fadd double [[TMP5]], [[TMP6]]
@@ -250,7 +250,7 @@ double f_va_2(char *fmt, ...) {
 // CHECK-ILP32E-NEXT:    [[W:%.*]] = alloca i32, align 4
 // CHECK-ILP32E-NEXT:    [[X:%.*]] = alloca double, align 8
 // CHECK-ILP32E-NEXT:    store ptr [[FMT]], ptr [[FMT_ADDR]], align 4
-// CHECK-ILP32E-NEXT:    call void @llvm.va_start(ptr [[VA]])
+// CHECK-ILP32E-NEXT:    call void @llvm.va_start.p0(ptr [[VA]])
 // CHECK-ILP32E-NEXT:    [[ARGP_CUR:%.*]] = load ptr, ptr [[VA]], align 4
 // CHECK-ILP32E-NEXT:    [[ARGP_NEXT:%.*]] = getelementptr inbounds i8, ptr [[ARGP_CUR]], i32 8
 // CHECK-ILP32E-NEXT:    store ptr [[ARGP_NEXT]], ptr [[VA]], align 4
@@ -266,7 +266,7 @@ double f_va_2(char *fmt, ...) {
 // CHECK-ILP32E-NEXT:    store ptr [[ARGP_NEXT4]], ptr [[VA]], align 4
 // CHECK-ILP32E-NEXT:    [[TMP2:%.*]] = load double, ptr [[ARGP_CUR3]], align 4
 // CHECK-ILP32E-NEXT:    store double [[TMP2]], ptr [[X]], align 8
-// CHECK-ILP32E-NEXT:    call void @llvm.va_end(ptr [[VA]])
+// CHECK-ILP32E-NEXT:    call void @llvm.va_end.p0(ptr [[VA]])
 // CHECK-ILP32E-NEXT:    [[TMP3:%.*]] = load double, ptr [[V]], align 8
 // CHECK-ILP32E-NEXT:    [[TMP4:%.*]] = load double, ptr [[X]], align 8
 // CHECK-ILP32E-NEXT:    [[ADD:%.*]] = fadd double [[TMP3]], [[TMP4]]
@@ -296,7 +296,7 @@ double f_va_3(char *fmt, ...) {
 // CHECK-ILP32F-NEXT:    [[LS:%.*]] = alloca [[STRUCT_LARGE:%.*]], align 4
 // CHECK-ILP32F-NEXT:    [[RET:%.*]] = alloca i32, align 4
 // CHECK-ILP32F-NEXT:    store ptr [[FMT]], ptr [[FMT_ADDR]], align 4
-// CHECK-ILP32F-NEXT:    call void @llvm.va_start(ptr [[VA]])
+// CHECK-ILP32F-NEXT:    call void @llvm.va_start.p0(ptr [[VA]])
 // CHECK-ILP32F-NEXT:    [[ARGP_CUR:%.*]] = load ptr, ptr [[VA]], align 4
 // CHECK-ILP32F-NEXT:    [[ARGP_NEXT:%.*]] = getelementptr inbounds i8, ptr [[ARGP_CUR]], i32 4
 // CHECK-ILP32F-NEXT:    store ptr [[ARGP_NEXT]], ptr [[VA]], align 4
@@ -321,7 +321,7 @@ double f_va_3(char *fmt, ...) {
 // CHECK-ILP32F-NEXT:    store ptr [[ARGP_NEXT8]], ptr [[VA]], align 4
 // CHECK-ILP32F-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[ARGP_CUR7]], align 4
 // CHECK-ILP32F-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[LS]], ptr align 4 [[TMP3]], i32 16, i1 false)
-// CHECK-ILP32F-NEXT:    call void @llvm.va_end(ptr [[VA]])
+// CHECK-ILP32F-NEXT:    call void @llvm.va_end.p0(ptr [[VA]])
 // CHECK-ILP32F-NEXT:    [[TMP4:%.*]] = load i32, ptr [[V]], align 4
 // CHECK-ILP32F-NEXT:    [[CONV:%.*]] = sitofp i32 [[TMP4]] to fp128
 // CHECK-ILP32F-NEXT:    [[TMP5:%.*]] = load fp128, ptr [[LD]], align 16
@@ -384,7 +384,7 @@ double f_va_3(char *fmt, ...) {
 // CHECK-ILP32D-NEXT:    [[LS:%.*]] = alloca [[STRUCT_LARGE:%.*]], align 4
 // CHECK-ILP32D-NEXT:    [[RET:%.*]] = alloca i32, align 4
 // CHECK-ILP32D-NEXT:    store ptr [[FMT]], ptr [[FMT_ADDR]], align 4
-// CHECK-ILP32D-NEXT:    call void @llvm.va_start(ptr [[VA]])
+// CHECK-ILP32D-NEXT:    call void @llvm.va_start.p0(ptr [[VA]])
 // CHECK-ILP32D-NEXT:    [[ARGP_CUR:%.*]] = load ptr, ptr [[VA]], align 4
 // CHECK-ILP32D-NEXT:    [[ARGP_NEXT:%.*]] = getelementptr inbounds i8, ptr [[ARGP_CUR]], i32 4
 // CHECK-ILP32D-NEXT:    store ptr [[ARGP_NEXT]], ptr [[VA]], align 4
@@ -409,7 +409,7 @@ double f_va_3(char *fmt, ...) {
 // CHECK-ILP32D-NEXT:    store ptr [[ARGP_NEXT8]], ptr [[VA]], align 4
 // CHECK-ILP32D-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[ARGP_CUR7]], align 4
 // CHECK-ILP32D-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[LS]], ptr align 4 [[TMP3]], i32 16, i1 false)
-// CHECK-ILP32D-NEXT:    call void @llvm.va_end(ptr [[VA]])
+// CHECK-ILP32D-NEXT:    call void @llvm.va_end.p0(ptr [[VA]])
 // CHECK-ILP32D-NEXT:    [[TMP4:%.*]] = load i32, ptr [[V]], align 4
 // CHECK-ILP32D-NEXT:    [[CONV:%.*]] = sitofp i32 [[TMP4]] to fp128
 // CHECK-ILP32D-NEXT:    [[TMP5:%.*]] = load fp128, ptr [[LD]], align 16
@@ -472,7 +472,7 @@ double f_va_3(char *fmt, ...) {
 // CHECK-ILP32E-NEXT:    [[LS:%.*]] = alloca [[STRUCT_LARGE:%.*]], align 4
 // CHECK-ILP32E-NEXT:    [[RET:%.*]] = alloca i32, align 4
 // CHECK-ILP32E-NEXT:    store ptr [[FMT]], ptr [[FMT_ADDR]], align 4
-// CHECK-ILP32E-NEXT:    call void @llvm.va_start(ptr [[VA]])
+// CHECK-ILP32E-NEXT:    call void @llvm.va_start.p0(ptr [[VA]])
 // CHECK-ILP32E-NEXT:    [[ARGP_CUR:%.*]] = load ptr, ptr [[VA]], align 4
 // CHECK-ILP32E-NEXT:    [[ARGP_NEXT:%.*]] = getelementptr inbounds i8, ptr [[ARGP_CUR]], i32 4
 // CHECK-ILP32E-NEXT:    store ptr [[ARGP_NEXT]], ptr [[VA]], align 4
@@ -497,7 +497,7 @@ double f_va_3(char *fmt, ...) {
 // CHECK-ILP32E-NEXT:    store ptr [[ARGP_NEXT8]], ptr [[VA]], align 4
 // CHECK-ILP32E-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[ARGP_CUR7]], align 4
 // CHECK-ILP32E-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[LS]], ptr align 4 [[TMP3]], i32 16, i1 false)
-// CHECK-ILP32E-NEXT:    call void @llvm.va_end(ptr [[VA]])
+// CHECK-ILP32E-NEXT:    call void @llvm.va_end.p0(ptr [[VA]])
 // CHECK-ILP32E-NEXT:    [[TMP4:%.*]] = load i32, ptr [[V]], align 4
 // CHECK-ILP32E-NEXT:    [[CONV:%.*]] = sitofp i32 [[TMP4]] to fp128
 // CHECK-ILP32E-NEXT:    [[TMP5:%.*]] = load fp128, ptr [[LD]], align 16

diff --git a/clang/test/CodeGen/RISCV/riscv64-vararg.c b/clang/test/CodeGen/RISCV/riscv64-vararg.c
@@ -135,13 +135,13 @@ void f_va_caller(void) {
 // CHECK-NEXT:    [[VA:%.*]] = alloca ptr, align 8
 // CHECK-NEXT:    [[V:%.*]] = alloca i32, align 4
 // CHECK-NEXT:    store ptr [[FMT]], ptr [[FMT_ADDR]], align 8
-// CHECK-NEXT:    call void @llvm.va_start(ptr [[VA]])
+// CHECK-NEXT:    call void @llvm.va_start.p0(ptr [[VA]])
 // CHECK-NEXT:    [[ARGP_CUR:%.*]] = load ptr, ptr [[VA]], align 8
 // CHECK-NEXT:    [[ARGP_NEXT:%.*]] = getelementptr inbounds i8, ptr [[ARGP_CUR]], i64 8
 // CHECK-NEXT:    store ptr [[ARGP_NEXT]], ptr [[VA]], align 8
 // CHECK-NEXT:    [[TMP0:%.*]] = load i32, ptr [[ARGP_CUR]], align 8
 // CHECK-NEXT:    store i32 [[TMP0]], ptr [[V]], align 4
-// CHECK-NEXT:    call void @llvm.va_end(ptr [[VA]])
+// CHECK-NEXT:    call void @llvm.va_end.p0(ptr [[VA]])
 // CHECK-NEXT:    [[TMP1:%.*]] = load i32, ptr [[V]], align 4
 // CHECK-NEXT:    ret i32 [[TMP1]]
 //
@@ -166,15 +166,15 @@ int f_va_1(char *fmt, ...) {
 // CHECK-NEXT:    [[VA:%.*]] = alloca ptr, align 8
 // CHECK-NEXT:    [[V:%.*]] = alloca fp128, align 16
 // CHECK-NEXT:    store ptr [[FMT]], ptr [[FMT_ADDR]], align 8
-// CHECK-NEXT:    call void @llvm.va_start(ptr [[VA]])
+// CHECK-NEXT:    call void @llvm.va_start.p0(ptr [[VA]])
 // CHECK-NEXT:    [[ARGP_CUR:%.*]] = load ptr, ptr [[VA]], align 8
 // CHECK-NEXT:    [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[ARGP_CUR]], i32 15
 // CHECK-NEXT:    [[ARGP_CUR_ALIGNED:%.*]] = call ptr @llvm.ptrmask.p0.i64(ptr [[TMP0]], i64 -16)
 // CHECK-NEXT:    [[ARGP_NEXT:%.*]] = getelementptr inbounds i8, ptr [[ARGP_CUR_ALIGNED]], i64 16
 // CHECK-NEXT:    store ptr [[ARGP_NEXT]], ptr [[VA]], align 8
 // CHECK-NEXT:    [[TMP1:%.*]] = load fp128, ptr [[ARGP_CUR_ALIGNED]], align 16
 // CHECK-NEXT:    store fp128 [[TMP1]], ptr [[V]], align 16
-// CHECK-NEXT:    call void @llvm.va_end(ptr [[VA]])
+// CHECK-NEXT:    call void @llvm.va_end.p0(ptr [[VA]])
 // CHECK-NEXT:    [[TMP2:%.*]] = load fp128, ptr [[V]], align 16
 // CHECK-NEXT:    ret fp128 [[TMP2]]
 //
@@ -199,7 +199,7 @@ long double f_va_2(char *fmt, ...) {
 // CHECK-NEXT:    [[W:%.*]] = alloca i32, align 4
 // CHECK-NEXT:    [[X:%.*]] = alloca fp128, align 16
 // CHECK-NEXT:    store ptr [[FMT]], ptr [[FMT_ADDR]], align 8
-// CHECK-NEXT:    call void @llvm.va_start(ptr [[VA]])
+// CHECK-NEXT:    call void @llvm.va_start.p0(ptr [[VA]])
 // CHECK-NEXT:    [[ARGP_CUR:%.*]] = load ptr, ptr [[VA]], align 8
 // CHECK-NEXT:    [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[ARGP_CUR]], i32 15
 // CHECK-NEXT:    [[ARGP_CUR_ALIGNED:%.*]] = call ptr @llvm.ptrmask.p0.i64(ptr [[TMP0]], i64 -16)
@@ -219,7 +219,7 @@ long double f_va_2(char *fmt, ...) {
 // CHECK-NEXT:    store ptr [[ARGP_NEXT4]], ptr [[VA]], align 8
 // CHECK-NEXT:    [[TMP4:%.*]] = load fp128, ptr [[ARGP_CUR3_ALIGNED]], align 16
 // CHECK-NEXT:    store fp128 [[TMP4]], ptr [[X]], align 16
-// CHECK-NEXT:    call void @llvm.va_end(ptr [[VA]])
+// CHECK-NEXT:    call void @llvm.va_end.p0(ptr [[VA]])
 // CHECK-NEXT:    [[TMP5:%.*]] = load fp128, ptr [[V]], align 16
 // CHECK-NEXT:    [[TMP6:%.*]] = load fp128, ptr [[X]], align 16
 // CHECK-NEXT:    [[ADD:%.*]] = fadd fp128 [[TMP5]], [[TMP6]]
@@ -248,7 +248,7 @@ long double f_va_3(char *fmt, ...) {
 // CHECK-NEXT:    [[LS:%.*]] = alloca [[STRUCT_LARGE:%.*]], align 8
 // CHECK-NEXT:    [[RET:%.*]] = alloca i32, align 4
 // CHECK-NEXT:    store ptr [[FMT]], ptr [[FMT_ADDR]], align 8
-// CHECK-NEXT:    call void @llvm.va_start(ptr [[VA]])
+// CHECK-NEXT:    call void @llvm.va_start.p0(ptr [[VA]])
 // CHECK-NEXT:    [[ARGP_CUR:%.*]] = load ptr, ptr [[VA]], align 8
 // CHECK-NEXT:    [[ARGP_NEXT:%.*]] = getelementptr inbounds i8, ptr [[ARGP_CUR]], i64 8
 // CHECK-NEXT:    store ptr [[ARGP_NEXT]], ptr [[VA]], align 8
@@ -267,7 +267,7 @@ long double f_va_3(char *fmt, ...) {
 // CHECK-NEXT:    store ptr [[ARGP_NEXT6]], ptr [[VA]], align 8
 // CHECK-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[ARGP_CUR5]], align 8
 // CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[LS]], ptr align 8 [[TMP1]], i64 32, i1 false)
-// CHECK-NEXT:    call void @llvm.va_end(ptr [[VA]])
+// CHECK-NEXT:    call void @llvm.va_end.p0(ptr [[VA]])
 // CHECK-NEXT:    [[A:%.*]] = getelementptr inbounds [[STRUCT_TINY]], ptr [[TS]], i32 0, i32 0
 // CHECK-NEXT:    [[TMP2:%.*]] = load i16, ptr [[A]], align 2
 // CHECK-NEXT:    [[CONV:%.*]] = zext i16 [[TMP2]] to i64

diff --git a/clang/test/CodeGen/RISCV/rvb-intrinsics/riscv32-zbb-error.c b/clang/test/CodeGen/RISCV/rvb-intrinsics/riscv32-zbb-error.c
@@ -1,6 +1,6 @@
 // NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
-// RUN: %clang_cc1 -triple riscv32 -target-feature +zbb -verify %s -o -
+// RUN: %clang_cc1 -triple riscv32 -target-feature +zbb -S -verify %s -o -
 
 unsigned int orc_b_64(unsigned int a) {
-  return __builtin_riscv_orc_b_64(a); // expected-error {{builtin requires: 'RV64'}}
+  return __builtin_riscv_orc_b_64(a); // expected-error {{'__builtin_riscv_orc_b_64' needs target feature zbb,64bit}}
 }
diff --git a/clang/test/CodeGen/RISCV/rvb-intrinsics/riscv64-zbkb-error.c b/clang/test/CodeGen/RISCV/rvb-intrinsics/riscv64-zbkb-error.c
@@ -1,14 +1,10 @@
 // NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
-// RUN: %clang_cc1 -triple riscv64 -target-feature +zbkb -verify %s -o -
+// RUN: %clang_cc1 -triple riscv64 -target-feature +zbkb -S -verify %s -o -
 
 #include <stdint.h>
 
-uint32_t zip(uint32_t rs1)
+uint32_t zip_unzip(uint32_t rs1)
 {
-  return __builtin_riscv_zip_32(rs1); // expected-error {{builtin requires: 'RV32'}}
-}
-
-uint32_t unzip(uint32_t rs1)
-{
-  return __builtin_riscv_unzip_32(rs1); // expected-error {{builtin requires: 'RV32'}}
+  (void)__builtin_riscv_zip_32(rs1); // expected-error {{'__builtin_riscv_zip_32' needs target feature zbkb,32bit}}
+  return __builtin_riscv_unzip_32(rs1); // expected-error {{'__builtin_riscv_unzip_32' needs target feature zbkb,32bit}}
 }
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-handcrafted/rvv-error.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-handcrafted/rvv-error.c
@@ -11,7 +11,7 @@
 // CHECK-RV64V-NEXT:    ret i32 [[CONV]]
 //
 
-// CHECK-RV64-ERR: error: builtin requires at least one of the following extensions: 'Zve32x'
+// CHECK-RV64-ERR: error: '__builtin_rvv_vsetvli' needs target feature zve32x
 
 int test() {
   return __builtin_rvv_vsetvli(1, 0, 0);

diff --git a/clang/test/CodeGen/WebAssembly/wasm-varargs.c b/clang/test/CodeGen/WebAssembly/wasm-varargs.c
@@ -10,13 +10,13 @@
 // CHECK-NEXT:    [[VA:%.*]] = alloca ptr, align 4
 // CHECK-NEXT:    [[V:%.*]] = alloca i32, align 4
 // CHECK-NEXT:    store ptr [[FMT]], ptr [[FMT_ADDR]], align 4
-// CHECK-NEXT:    call void @llvm.va_start(ptr [[VA]])
+// CHECK-NEXT:    call void @llvm.va_start.p0(ptr [[VA]])
 // CHECK-NEXT:    [[ARGP_CUR:%.*]] = load ptr, ptr [[VA]], align 4
 // CHECK-NEXT:    [[ARGP_NEXT:%.*]] = getelementptr inbounds i8, ptr [[ARGP_CUR]], i32 4
 // CHECK-NEXT:    store ptr [[ARGP_NEXT]], ptr [[VA]], align 4
 // CHECK-NEXT:    [[TMP0:%.*]] = load i32, ptr [[ARGP_CUR]], align 4
 // CHECK-NEXT:    store i32 [[TMP0]], ptr [[V]], align 4
-// CHECK-NEXT:    call void @llvm.va_end(ptr [[VA]])
+// CHECK-NEXT:    call void @llvm.va_end.p0(ptr [[VA]])
 // CHECK-NEXT:    [[TMP1:%.*]] = load i32, ptr [[V]], align 4
 // CHECK-NEXT:    ret i32 [[TMP1]]
 //
@@ -38,15 +38,15 @@ int test_i32(char *fmt, ...) {
 // CHECK-NEXT:    [[VA:%.*]] = alloca ptr, align 4
 // CHECK-NEXT:    [[V:%.*]] = alloca i64, align 8
 // CHECK-NEXT:    store ptr [[FMT]], ptr [[FMT_ADDR]], align 4
-// CHECK-NEXT:    call void @llvm.va_start(ptr [[VA]])
+// CHECK-NEXT:    call void @llvm.va_start.p0(ptr [[VA]])
 // CHECK-NEXT:    [[ARGP_CUR:%.*]] = load ptr, ptr [[VA]], align 4
 // CHECK-NEXT:    [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[ARGP_CUR]], i32 7
 // CHECK-NEXT:    [[ARGP_CUR_ALIGNED:%.*]] = call ptr @llvm.ptrmask.p0.i32(ptr [[TMP0]], i32 -8)
 // CHECK-NEXT:    [[ARGP_NEXT:%.*]] = getelementptr inbounds i8, ptr [[ARGP_CUR_ALIGNED]], i32 8
 // CHECK-NEXT:    store ptr [[ARGP_NEXT]], ptr [[VA]], align 4
 // CHECK-NEXT:    [[TMP1:%.*]] = load i64, ptr [[ARGP_CUR_ALIGNED]], align 8
 // CHECK-NEXT:    store i64 [[TMP1]], ptr [[V]], align 8
-// CHECK-NEXT:    call void @llvm.va_end(ptr [[VA]])
+// CHECK-NEXT:    call void @llvm.va_end.p0(ptr [[VA]])
 // CHECK-NEXT:    [[TMP2:%.*]] = load i64, ptr [[V]], align 8
 // CHECK-NEXT:    ret i64 [[TMP2]]
 //
@@ -73,13 +73,13 @@ struct S {
 // CHECK-NEXT:    [[FMT_ADDR:%.*]] = alloca ptr, align 4
 // CHECK-NEXT:    [[VA:%.*]] = alloca ptr, align 4
 // CHECK-NEXT:    store ptr [[FMT]], ptr [[FMT_ADDR]], align 4
-// CHECK-NEXT:    call void @llvm.va_start(ptr [[VA]])
+// CHECK-NEXT:    call void @llvm.va_start.p0(ptr [[VA]])
 // CHECK-NEXT:    [[ARGP_CUR:%.*]] = load ptr, ptr [[VA]], align 4
 // CHECK-NEXT:    [[ARGP_NEXT:%.*]] = getelementptr inbounds i8, ptr [[ARGP_CUR]], i32 4
 // CHECK-NEXT:    store ptr [[ARGP_NEXT]], ptr [[VA]], align 4
 // CHECK-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[ARGP_CUR]], align 4
 // CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[AGG_RESULT]], ptr align 4 [[TMP0]], i32 12, i1 false)
-// CHECK-NEXT:    call void @llvm.va_end(ptr [[VA]])
+// CHECK-NEXT:    call void @llvm.va_end.p0(ptr [[VA]])
 // CHECK-NEXT:    ret void
 //
 struct S test_struct(char *fmt, ...) {
@@ -102,7 +102,7 @@ struct Z {};
 // CHECK-NEXT:    [[VA:%.*]] = alloca ptr, align 4
 // CHECK-NEXT:    [[U:%.*]] = alloca [[STRUCT_Z:%.*]], align 1
 // CHECK-NEXT:    store ptr [[FMT]], ptr [[FMT_ADDR]], align 4
-// CHECK-NEXT:    call void @llvm.va_start(ptr [[VA]])
+// CHECK-NEXT:    call void @llvm.va_start.p0(ptr [[VA]])
 // CHECK-NEXT:    [[ARGP_CUR:%.*]] = load ptr, ptr [[VA]], align 4
 // CHECK-NEXT:    [[ARGP_NEXT:%.*]] = getelementptr inbounds i8, ptr [[ARGP_CUR]], i32 0
 // CHECK-NEXT:    store ptr [[ARGP_NEXT]], ptr [[VA]], align 4
@@ -112,7 +112,7 @@ struct Z {};
 // CHECK-NEXT:    store ptr [[ARGP_NEXT2]], ptr [[VA]], align 4
 // CHECK-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[ARGP_CUR1]], align 4
 // CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[AGG_RESULT]], ptr align 4 [[TMP0]], i32 12, i1 false)
-// CHECK-NEXT:    call void @llvm.va_end(ptr [[VA]])
+// CHECK-NEXT:    call void @llvm.va_end.p0(ptr [[VA]])
 // CHECK-NEXT:    ret void
 //
 struct S test_empty_struct(char *fmt, ...) {

diff --git a/clang/test/CodeGen/X86/va-arg-sse.c b/clang/test/CodeGen/X86/va-arg-sse.c
@@ -21,7 +21,7 @@ struct S a[5];
 // CHECK-NEXT:    store i32 0, ptr [[J]], align 4
 // CHECK-NEXT:    store i32 0, ptr [[K]], align 4
 // CHECK-NEXT:    [[ARRAYDECAY:%.*]] = getelementptr inbounds [1 x %struct.__va_list_tag], ptr [[AP]], i64 0, i64 0
-// CHECK-NEXT:    call void @llvm.va_start(ptr [[ARRAYDECAY]])
+// CHECK-NEXT:    call void @llvm.va_start.p0(ptr [[ARRAYDECAY]])
 // CHECK-NEXT:    store ptr getelementptr inbounds ([5 x %struct.S], ptr @a, i64 0, i64 2), ptr [[P]], align 8
 // CHECK-NEXT:    [[ARRAYDECAY2:%.*]] = getelementptr inbounds [1 x %struct.__va_list_tag], ptr [[AP]], i64 0, i64 0
 // CHECK-NEXT:    [[FP_OFFSET_P:%.*]] = getelementptr inbounds [[STRUCT___VA_LIST_TAG:%.*]], ptr [[ARRAYDECAY2]], i32 0, i32 1
@@ -52,7 +52,7 @@ struct S a[5];
 // CHECK-NEXT:    [[VAARG_ADDR:%.*]] = phi ptr [ [[TMP]], [[VAARG_IN_REG]] ], [ [[OVERFLOW_ARG_AREA]], [[VAARG_IN_MEM]] ]
 // CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[ARG]], ptr align 4 [[VAARG_ADDR]], i64 12, i1 false)
 // CHECK-NEXT:    [[ARRAYDECAY3:%.*]] = getelementptr inbounds [1 x %struct.__va_list_tag], ptr [[AP]], i64 0, i64 0
-// CHECK-NEXT:    call void @llvm.va_end(ptr [[ARRAYDECAY3]])
+// CHECK-NEXT:    call void @llvm.va_end.p0(ptr [[ARRAYDECAY3]])
 // CHECK-NEXT:    [[TMP15:%.*]] = load ptr, ptr [[P]], align 8
 // CHECK-NEXT:    [[TOBOOL:%.*]] = icmp ne ptr [[TMP15]], null
 // CHECK-NEXT:    br i1 [[TOBOOL]], label [[LAND_LHS_TRUE:%.*]], label [[IF_END:%.*]]

diff --git a/clang/test/CodeGen/X86/x86_64-vaarg.c b/clang/test/CodeGen/X86/x86_64-vaarg.c
@@ -13,7 +13,7 @@ typedef struct { struct {} a; } empty;
 // CHECK-NEXT:    [[TMP:%.*]] = alloca [[STRUCT_EMPTY]], align 1
 // CHECK-NEXT:    store i32 [[Z]], ptr [[Z_ADDR]], align 4
 // CHECK-NEXT:    [[ARRAYDECAY:%.*]] = getelementptr inbounds [1 x %struct.__va_list_tag], ptr [[LIST]], i64 0, i64 0
-// CHECK-NEXT:    call void @llvm.va_start(ptr [[ARRAYDECAY]])
+// CHECK-NEXT:    call void @llvm.va_start.p0(ptr [[ARRAYDECAY]])
 // CHECK-NEXT:    [[ARRAYDECAY1:%.*]] = getelementptr inbounds [1 x %struct.__va_list_tag], ptr [[LIST]], i64 0, i64 0
 // CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 1 [[RETVAL]], ptr align 1 [[TMP]], i64 0, i1 false)
 // CHECK-NEXT:    ret void
@@ -37,7 +37,7 @@ typedef struct {
 // CHECK-NEXT:    [[LIST:%.*]] = alloca [1 x %struct.__va_list_tag], align 16
 // CHECK-NEXT:    store i32 [[Z]], ptr [[Z_ADDR]], align 4
 // CHECK-NEXT:    [[ARRAYDECAY:%.*]] = getelementptr inbounds [1 x %struct.__va_list_tag], ptr [[LIST]], i64 0, i64 0
-// CHECK-NEXT:    call void @llvm.va_start(ptr [[ARRAYDECAY]])
+// CHECK-NEXT:    call void @llvm.va_start.p0(ptr [[ARRAYDECAY]])
 // CHECK-NEXT:    [[ARRAYDECAY1:%.*]] = getelementptr inbounds [1 x %struct.__va_list_tag], ptr [[LIST]], i64 0, i64 0
 // CHECK-NEXT:    [[FP_OFFSET_P:%.*]] = getelementptr inbounds [[STRUCT___VA_LIST_TAG:%.*]], ptr [[ARRAYDECAY1]], i32 0, i32 1
 // CHECK-NEXT:    [[FP_OFFSET:%.*]] = load i32, ptr [[FP_OFFSET_P]], align 4

diff --git a/clang/test/CodeGen/aarch64-ABI-align-packed.c b/clang/test/CodeGen/aarch64-ABI-align-packed.c
@@ -73,7 +73,7 @@ __attribute__((noinline)) void named_arg_non_packed_struct(double d0, double d1,
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[VL:%.*]] = alloca [[STRUCT___VA_LIST:%.*]], align 8
 // CHECK-NEXT:    call void @llvm.lifetime.start.p0(i64 32, ptr nonnull [[VL]]) #[[ATTR6:[0-9]+]]
-// CHECK-NEXT:    call void @llvm.va_start(ptr nonnull [[VL]])
+// CHECK-NEXT:    call void @llvm.va_start.p0(ptr nonnull [[VL]])
 // CHECK-NEXT:    call void @llvm.lifetime.end.p0(i64 32, ptr nonnull [[VL]]) #[[ATTR6]]
 // CHECK-NEXT:    ret void
 void variadic_non_packed_struct(double d0, double d1, double d2, double d3,
@@ -128,7 +128,7 @@ __attribute__((noinline)) void named_arg_packed_struct(double d0, double d1, dou
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[VL:%.*]] = alloca [[STRUCT___VA_LIST:%.*]], align 8
 // CHECK-NEXT:    call void @llvm.lifetime.start.p0(i64 32, ptr nonnull [[VL]]) #[[ATTR6]]
-// CHECK-NEXT:    call void @llvm.va_start(ptr nonnull [[VL]])
+// CHECK-NEXT:    call void @llvm.va_start.p0(ptr nonnull [[VL]])
 // CHECK-NEXT:    call void @llvm.lifetime.end.p0(i64 32, ptr nonnull [[VL]]) #[[ATTR6]]
 // CHECK-NEXT:    ret void
 void variadic_packed_struct(double d0, double d1, double d2, double d3,
@@ -183,7 +183,7 @@ __attribute__((noinline)) void named_arg_packed_member(double d0, double d1, dou
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[VL:%.*]] = alloca [[STRUCT___VA_LIST:%.*]], align 8
 // CHECK-NEXT:    call void @llvm.lifetime.start.p0(i64 32, ptr nonnull [[VL]]) #[[ATTR6]]
-// CHECK-NEXT:    call void @llvm.va_start(ptr nonnull [[VL]])
+// CHECK-NEXT:    call void @llvm.va_start.p0(ptr nonnull [[VL]])
 // CHECK-NEXT:    call void @llvm.lifetime.end.p0(i64 32, ptr nonnull [[VL]]) #[[ATTR6]]
 // CHECK-NEXT:    ret void
 void variadic_packed_member(double d0, double d1, double d2, double d3,
@@ -238,7 +238,7 @@ __attribute__((noinline)) void named_arg_aligned_struct_8(double d0, double d1,
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[VL:%.*]] = alloca [[STRUCT___VA_LIST:%.*]], align 8
 // CHECK-NEXT:    call void @llvm.lifetime.start.p0(i64 32, ptr nonnull [[VL]]) #[[ATTR6]]
-// CHECK-NEXT:    call void @llvm.va_start(ptr nonnull [[VL]])
+// CHECK-NEXT:    call void @llvm.va_start.p0(ptr nonnull [[VL]])
 // CHECK-NEXT:    call void @llvm.lifetime.end.p0(i64 32, ptr nonnull [[VL]]) #[[ATTR6]]
 // CHECK-NEXT:    ret void
 void variadic_aligned_struct_8(double d0, double d1, double d2, double d3,
@@ -293,7 +293,7 @@ __attribute__((noinline)) void named_arg_aligned_member_8(double d0, double d1,
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[VL:%.*]] = alloca [[STRUCT___VA_LIST:%.*]], align 8
 // CHECK-NEXT:    call void @llvm.lifetime.start.p0(i64 32, ptr nonnull [[VL]]) #[[ATTR6]]
-// CHECK-NEXT:    call void @llvm.va_start(ptr nonnull [[VL]])
+// CHECK-NEXT:    call void @llvm.va_start.p0(ptr nonnull [[VL]])
 // CHECK-NEXT:    call void @llvm.lifetime.end.p0(i64 32, ptr nonnull [[VL]]) #[[ATTR6]]
 // CHECK-NEXT:    ret void
 void variadic_aligned_member_8(double d0, double d1, double d2, double d3,
@@ -348,7 +348,7 @@ __attribute__((noinline)) void named_arg_pragma_packed_struct_8(double d0, doubl
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[VL:%.*]] = alloca [[STRUCT___VA_LIST:%.*]], align 8
 // CHECK-NEXT:    call void @llvm.lifetime.start.p0(i64 32, ptr nonnull [[VL]]) #[[ATTR6]]
-// CHECK-NEXT:    call void @llvm.va_start(ptr nonnull [[VL]])
+// CHECK-NEXT:    call void @llvm.va_start.p0(ptr nonnull [[VL]])
 // CHECK-NEXT:    call void @llvm.lifetime.end.p0(i64 32, ptr nonnull [[VL]]) #[[ATTR6]]
 // CHECK-NEXT:    ret void
 void variadic_pragma_packed_struct_8(double d0, double d1, double d2, double d3,
@@ -403,7 +403,7 @@ __attribute__((noinline)) void named_arg_pragma_packed_struct_4(double d0, doubl
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[VL:%.*]] = alloca [[STRUCT___VA_LIST:%.*]], align 8
 // CHECK-NEXT:    call void @llvm.lifetime.start.p0(i64 32, ptr nonnull [[VL]]) #[[ATTR6]]
-// CHECK-NEXT:    call void @llvm.va_start(ptr nonnull [[VL]])
+// CHECK-NEXT:    call void @llvm.va_start.p0(ptr nonnull [[VL]])
 // CHECK-NEXT:    call void @llvm.lifetime.end.p0(i64 32, ptr nonnull [[VL]]) #[[ATTR6]]
 // CHECK-NEXT:    ret void
 void variadic_pragma_packed_struct_4(double d0, double d1, double d2, double d3,

diff --git a/clang/test/CodeGen/aarch64-varargs.c b/clang/test/CodeGen/aarch64-varargs.c
@@ -837,7 +837,7 @@ void check_start(int n, ...) {
   va_list the_list;
   va_start(the_list, n);
 // CHECK: [[THE_LIST:%[a-z_0-9]+]] = alloca %struct.__va_list
-// CHECK: call void @llvm.va_start(ptr [[THE_LIST]])
+// CHECK: call void @llvm.va_start.p0(ptr [[THE_LIST]])
 }
 
 typedef struct {} empty;

diff --git a/clang/test/CodeGen/arm-varargs.c b/clang/test/CodeGen/arm-varargs.c
@@ -264,5 +264,5 @@ void check_start(int n, ...) {
   va_list the_list;
   va_start(the_list, n);
 // CHECK: [[THE_LIST:%[a-z0-9._]+]] = alloca %struct.__va_list
-// CHECK: call void @llvm.va_start(ptr [[THE_LIST]])
+// CHECK: call void @llvm.va_start.p0(ptr [[THE_LIST]])
 }
diff --git a/clang/test/CodeGen/hexagon-linux-vararg.c b/clang/test/CodeGen/hexagon-linux-vararg.c
@@ -9,7 +9,7 @@ struct AAA {
   int d;
 };
 
-// CHECK:   call void @llvm.va_start(ptr %arraydecay)
+// CHECK:   call void @llvm.va_start.p0(ptr %arraydecay)
 // CHECK:   %arraydecay1 = getelementptr inbounds [1 x %struct.__va_list_tag],
 // ptr %ap, i32 0, i32 0
 // CHECK:   br label %vaarg.maybe_reg

diff --git a/clang/test/CodeGen/mips-varargs.c b/clang/test/CodeGen/mips-varargs.c
@@ -29,7 +29,7 @@ int test_i32(char *fmt, ...) {
 // ALL:   [[V:%.*]] = alloca i32, align 4
 // NEW:   [[PROMOTION_TEMP:%.*]] = alloca i32, align 4
 //
-// ALL:   call void @llvm.va_start(ptr %va)
+// ALL:   call void @llvm.va_start.p0(ptr %va)
 // ALL:   [[AP_CUR:%.+]] = load ptr, ptr %va, align [[$PTRALIGN]]
 // O32:   [[AP_NEXT:%.+]] = getelementptr inbounds i8, ptr [[AP_CUR]], [[$INTPTR_T:i32]] [[$CHUNKSIZE:4]]
 // NEW:   [[AP_NEXT:%.+]] = getelementptr inbounds i8, ptr [[AP_CUR]], [[$INTPTR_T:i32|i64]] [[$CHUNKSIZE:8]]
@@ -45,7 +45,7 @@ int test_i32(char *fmt, ...) {
 // NEW:   [[ARG:%.+]] = load i32, ptr [[PROMOTION_TEMP]], align 4
 // ALL:   store i32 [[ARG]], ptr [[V]], align 4
 //
-// ALL:   call void @llvm.va_end(ptr %va)
+// ALL:   call void @llvm.va_end.p0(ptr %va)
 // ALL: }
 
 long long test_i64(char *fmt, ...) {
@@ -61,7 +61,7 @@ long long test_i64(char *fmt, ...) {
 // ALL-LABEL: define{{.*}} i64 @test_i64(ptr{{.*}} %fmt, ...)
 //
 // ALL:   %va = alloca ptr, align [[$PTRALIGN]]
-// ALL:   call void @llvm.va_start(ptr %va)
+// ALL:   call void @llvm.va_start.p0(ptr %va)
 // ALL:   [[AP_CUR:%.+]] = load ptr, ptr %va, align [[$PTRALIGN]]
 //
 // i64 is 8-byte aligned, while this is within O32's stack alignment there's no
@@ -74,7 +74,7 @@ long long test_i64(char *fmt, ...) {
 //
 // ALL:   [[ARG:%.+]] = load i64, ptr [[AP_CUR]], align 8
 //
-// ALL:   call void @llvm.va_end(ptr %va)
+// ALL:   call void @llvm.va_end.p0(ptr %va)
 // ALL: }
 
 char *test_ptr(char *fmt, ...) {
@@ -92,7 +92,7 @@ char *test_ptr(char *fmt, ...) {
 // ALL:   %va = alloca ptr, align [[$PTRALIGN]]
 // ALL:   [[V:%.*]] = alloca ptr, align [[$PTRALIGN]]
 // N32:   [[AP_CAST:%.+]] = alloca ptr, align 4
-// ALL:   call void @llvm.va_start(ptr %va)
+// ALL:   call void @llvm.va_start.p0(ptr %va)
 // ALL:   [[AP_CUR:%.+]] = load ptr, ptr %va, align [[$PTRALIGN]]
 // ALL:   [[AP_NEXT:%.+]] = getelementptr inbounds i8, ptr [[AP_CUR]], [[$INTPTR_T]] [[$CHUNKSIZE]]
 // ALL:   store ptr [[AP_NEXT]], ptr %va, align [[$PTRALIGN]]
@@ -109,7 +109,7 @@ char *test_ptr(char *fmt, ...) {
 // N64:   [[ARG:%.+]] = load ptr, ptr [[AP_CUR]], align [[$PTRALIGN]]
 // ALL:   store ptr [[ARG]], ptr [[V]], align [[$PTRALIGN]]
 //
-// ALL:   call void @llvm.va_end(ptr %va)
+// ALL:   call void @llvm.va_end.p0(ptr %va)
 // ALL: }
 
 int test_v4i32(char *fmt, ...) {
@@ -128,7 +128,7 @@ int test_v4i32(char *fmt, ...) {
 //
 // ALL:   %va = alloca ptr, align [[$PTRALIGN]]
 // ALL:   [[V:%.+]] = alloca <4 x i32>, align 16
-// ALL:   call void @llvm.va_start(ptr %va)
+// ALL:   call void @llvm.va_start.p0(ptr %va)
 // ALL:   [[AP_CUR:%.+]] = load ptr, ptr %va, align [[$PTRALIGN]]
 //
 // Vectors are 16-byte aligned, however the O32 ABI has a maximum alignment of
@@ -152,7 +152,7 @@ int test_v4i32(char *fmt, ...) {
 // N32:   [[ARG:%.+]] = load <4 x i32>, ptr [[AP_CUR]], align 16
 // ALL:   store <4 x i32> [[ARG]], ptr [[V]], align 16
 //
-// ALL:   call void @llvm.va_end(ptr %va)
+// ALL:   call void @llvm.va_end.p0(ptr %va)
 // ALL:   [[VECEXT:%.+]] = extractelement <4 x i32> {{.*}}, i32 0
 // ALL:   ret i32 [[VECEXT]]
 // ALL: }
diff --git a/clang/test/CodeGen/pr53127.cpp b/clang/test/CodeGen/pr53127.cpp
@@ -34,7 +34,7 @@ void operator delete(void*);
 // CHECK-NEXT:    br i1 [[CALL6]], label [[COND_TRUE7:%.*]], label [[COND_FALSE8:%.*]]
 // CHECK:       cond.true7:
 // CHECK-NEXT:    [[ARRAYDECAY:%.*]] = getelementptr inbounds [1 x %struct.__va_list_tag], ptr [[L]], i64 0, i64 0
-// CHECK-NEXT:    call void @llvm.va_start(ptr [[ARRAYDECAY]])
+// CHECK-NEXT:    call void @llvm.va_start.p0(ptr [[ARRAYDECAY]])
 // CHECK-NEXT:    br label [[COND_END9:%.*]]
 // CHECK:       cond.false8:
 // CHECK-NEXT:    br label [[COND_END9]]
@@ -44,7 +44,7 @@ void operator delete(void*);
 // CHECK:       cond.true11:
 // CHECK-NEXT:    [[ARRAYDECAY12:%.*]] = getelementptr inbounds [1 x %struct.__va_list_tag], ptr [[L]], i64 0, i64 0
 // CHECK-NEXT:    [[ARRAYDECAY13:%.*]] = getelementptr inbounds [1 x %struct.__va_list_tag], ptr [[L2]], i64 0, i64 0
-// CHECK-NEXT:    call void @llvm.va_copy(ptr [[ARRAYDECAY12]], ptr [[ARRAYDECAY13]])
+// CHECK-NEXT:    call void @llvm.va_copy.p0(ptr [[ARRAYDECAY12]], ptr [[ARRAYDECAY13]])
 // CHECK-NEXT:    br label [[COND_END15:%.*]]
 // CHECK:       cond.false14:
 // CHECK-NEXT:    br label [[COND_END15]]

diff --git a/clang/test/CodeGen/varargs-with-nonzero-default-address-space.c b/clang/test/CodeGen/varargs-with-nonzero-default-address-space.c
@@ -0,0 +1,46 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 4
+// RUN: %clang_cc1 -triple spirv64-unknown-unknown -fcuda-is-device -emit-llvm -o - %s | FileCheck %s
+
+struct x {
+  double b;
+  long a;
+};
+
+// CHECK-LABEL: define spir_func void @testva(
+// CHECK-SAME: i32 noundef [[N:%.*]], ...) #[[ATTR0:[0-9]+]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[N_ADDR:%.*]] = alloca i32, align 4
+// CHECK-NEXT:    [[AP:%.*]] = alloca ptr addrspace(4), align 8
+// CHECK-NEXT:    [[T:%.*]] = alloca [[STRUCT_X:%.*]], align 8
+// CHECK-NEXT:    [[AP2:%.*]] = alloca ptr addrspace(4), align 8
+// CHECK-NEXT:    [[V:%.*]] = alloca i32, align 4
+// CHECK-NEXT:    [[VARET:%.*]] = alloca i32, align 4
+// CHECK-NEXT:    [[N_ADDR_ASCAST:%.*]] = addrspacecast ptr [[N_ADDR]] to ptr addrspace(4)
+// CHECK-NEXT:    [[AP_ASCAST:%.*]] = addrspacecast ptr [[AP]] to ptr addrspace(4)
+// CHECK-NEXT:    [[T_ASCAST:%.*]] = addrspacecast ptr [[T]] to ptr addrspace(4)
+// CHECK-NEXT:    [[AP2_ASCAST:%.*]] = addrspacecast ptr [[AP2]] to ptr addrspace(4)
+// CHECK-NEXT:    [[V_ASCAST:%.*]] = addrspacecast ptr [[V]] to ptr addrspace(4)
+// CHECK-NEXT:    [[VARET_ASCAST:%.*]] = addrspacecast ptr [[VARET]] to ptr addrspace(4)
+// CHECK-NEXT:    store i32 [[N]], ptr addrspace(4) [[N_ADDR_ASCAST]], align 4
+// CHECK-NEXT:    call void @llvm.va_start.p4(ptr addrspace(4) [[AP_ASCAST]])
+// CHECK-NEXT:    [[TMP0:%.*]] = va_arg ptr addrspace(4) [[AP_ASCAST]], ptr
+// CHECK-NEXT:    call void @llvm.memcpy.p4.p0.i64(ptr addrspace(4) align 8 [[T_ASCAST]], ptr align 8 [[TMP0]], i64 16, i1 false)
+// CHECK-NEXT:    call void @llvm.va_copy.p4(ptr addrspace(4) [[AP2_ASCAST]], ptr addrspace(4) [[AP_ASCAST]])
+// CHECK-NEXT:    [[TMP1:%.*]] = va_arg ptr addrspace(4) [[AP2_ASCAST]], i32
+// CHECK-NEXT:    store i32 [[TMP1]], ptr addrspace(4) [[VARET_ASCAST]], align 4
+// CHECK-NEXT:    [[TMP2:%.*]] = load i32, ptr addrspace(4) [[VARET_ASCAST]], align 4
+// CHECK-NEXT:    store i32 [[TMP2]], ptr addrspace(4) [[V_ASCAST]], align 4
+// CHECK-NEXT:    call void @llvm.va_end.p4(ptr addrspace(4) [[AP2_ASCAST]])
+// CHECK-NEXT:    call void @llvm.va_end.p4(ptr addrspace(4) [[AP_ASCAST]])
+// CHECK-NEXT:    ret void
+
+void testva(int n, ...) {
+  __builtin_va_list ap;
+  __builtin_va_start(ap, n);
+  struct x t = __builtin_va_arg(ap, struct x);
+  __builtin_va_list ap2;
+  __builtin_va_copy(ap2, ap);
+  int v = __builtin_va_arg(ap2, int);
+  __builtin_va_end(ap2);
+  __builtin_va_end(ap);
+}
diff --git a/clang/test/CodeGen/xcore-abi.c b/clang/test/CodeGen/xcore-abi.c
@@ -28,7 +28,7 @@ void testva (int n, ...) {
   // CHECK: [[AP:%[a-z0-9]+]] = alloca ptr, align 4
   // CHECK: [[V5:%[a-z0-9]+]] = alloca %struct.x, align 4
   // CHECK: [[TMP:%[a-z0-9]+]] = alloca [4 x i32], align 4
-  // CHECK: call void @llvm.va_start(ptr [[AP]])
+  // CHECK: call void @llvm.va_start.p0(ptr [[AP]])
 
   char* v1 = va_arg (ap, char*);
   f(v1);

diff --git a/clang/test/CodeGenCXX/ext-int.cpp b/clang/test/CodeGenCXX/ext-int.cpp
@@ -159,9 +159,9 @@ void TakesVarargs(int i, ...) {
   // WIN: %[[ARGS:.+]] = alloca ptr
   __builtin_va_start(args, i);
   // LIN64: %[[STARTAD:.+]] = getelementptr inbounds [1 x %struct.__va_list_tag], ptr %[[ARGS]]
-  // LIN64: call void @llvm.va_start(ptr %[[STARTAD]])
-  // LIN32: call void @llvm.va_start(ptr %[[ARGS]])
-  // WIN: call void @llvm.va_start(ptr %[[ARGS]])
+  // LIN64: call void @llvm.va_start.p0(ptr %[[STARTAD]])
+  // LIN32: call void @llvm.va_start.p0(ptr %[[ARGS]])
+  // WIN: call void @llvm.va_start.p0(ptr %[[ARGS]])
 
   _BitInt(92) A = __builtin_va_arg(args, _BitInt(92));
   // LIN64: %[[AD1:.+]] = getelementptr inbounds [1 x %struct.__va_list_tag], ptr %[[ARGS]]
@@ -302,9 +302,9 @@ void TakesVarargs(int i, ...) {
 
   __builtin_va_end(args);
   // LIN64: %[[ENDAD:.+]] = getelementptr inbounds [1 x %struct.__va_list_tag], ptr %[[ARGS]]
-  // LIN64: call void @llvm.va_end(ptr %[[ENDAD]])
-  // LIN32: call void @llvm.va_end(ptr %[[ARGS]])
-  // WIN: call void @llvm.va_end(ptr %[[ARGS]])
+  // LIN64: call void @llvm.va_end.p0(ptr %[[ENDAD]])
+  // LIN32: call void @llvm.va_end.p0(ptr %[[ARGS]])
+  // WIN: call void @llvm.va_end.p0(ptr %[[ARGS]])
 }
 void typeid_tests() {
   // LIN: define{{.*}} void @_Z12typeid_testsv()

diff --git a/clang/test/CodeGenCXX/ibm128-declarations.cpp b/clang/test/CodeGenCXX/ibm128-declarations.cpp
@@ -107,13 +107,13 @@ int main(void) {
 // CHECK: define dso_local noundef ppc_fp128 @_Z10func_vaargiz(i32 noundef signext %n, ...)
 // CHECK: entry:
 // CHECK:   store i32 %n, ptr %n.addr, align 4
-// CHECK:   call void @llvm.va_start(ptr %ap)
+// CHECK:   call void @llvm.va_start.p0(ptr %ap)
 // CHECK:   %argp.cur = load ptr, ptr %ap, align 8
 // CHECK:   %argp.next = getelementptr inbounds i8, ptr %argp.cur, i64 16
 // CHECK:   store ptr %argp.next, ptr %ap, align 8
 // CHECK:   %0 = load ppc_fp128, ptr %argp.cur, align 8
 // CHECK:   store ppc_fp128 %0, ptr %r, align 16
-// CHECK:   call void @llvm.va_end(ptr %ap)
+// CHECK:   call void @llvm.va_end.p0(ptr %ap)
 // CHECK:   %1 = load ppc_fp128, ptr %r, align 16
 // CHECK:   ret ppc_fp128 %1
 // CHECK: }

diff --git a/clang/test/CodeGenCXX/x86_64-vaarg.cpp b/clang/test/CodeGenCXX/x86_64-vaarg.cpp
@@ -11,7 +11,7 @@ typedef struct { struct {} a; } empty;
 // CHECK-NEXT:    [[TMP:%.*]] = alloca [[STRUCT_EMPTY]], align 1
 // CHECK-NEXT:    store i32 [[Z:%.*]], ptr [[Z_ADDR]], align 4
 // CHECK-NEXT:    [[ARRAYDECAY:%.*]] = getelementptr inbounds [1 x %struct.__va_list_tag], ptr [[LIST]], i64 0, i64 0
-// CHECK-NEXT:    call void @llvm.va_start(ptr [[ARRAYDECAY]])
+// CHECK-NEXT:    call void @llvm.va_start.p0(ptr [[ARRAYDECAY]])
 // CHECK-NEXT:    [[ARRAYDECAY1:%.*]] = getelementptr inbounds [1 x %struct.__va_list_tag], ptr [[LIST]], i64 0, i64 0
 // CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 1 [[RETVAL]], ptr align 1 [[TMP]], i64 1, i1 false)
 // CHECK-NEXT:    ret void
@@ -34,7 +34,7 @@ typedef struct {
 // CHECK-NEXT:    [[LIST:%.*]] = alloca [1 x %struct.__va_list_tag], align 16
 // CHECK-NEXT:    store i32 [[Z:%.*]], ptr [[Z_ADDR]], align 4
 // CHECK-NEXT:    [[ARRAYDECAY:%.*]] = getelementptr inbounds [1 x %struct.__va_list_tag], ptr [[LIST]], i64 0, i64 0
-// CHECK-NEXT:    call void @llvm.va_start(ptr [[ARRAYDECAY]])
+// CHECK-NEXT:    call void @llvm.va_start.p0(ptr [[ARRAYDECAY]])
 // CHECK-NEXT:    [[ARRAYDECAY1:%.*]] = getelementptr inbounds [1 x %struct.__va_list_tag], ptr [[LIST]], i64 0, i64 0
 // CHECK-NEXT:    [[FP_OFFSET_P:%.*]] = getelementptr inbounds [[STRUCT___VA_LIST_TAG:%.*]], ptr [[ARRAYDECAY1]], i32 0, i32 1
 // CHECK-NEXT:    [[FP_OFFSET:%.*]] = load i32, ptr [[FP_OFFSET_P]], align 4

diff --git a/clang/test/Driver/aarch64-sve.c b/clang/test/Driver/aarch64-sve.c
@@ -6,12 +6,11 @@
 // RUN: %clang --target=aarch64 -march=armv8.6a -### -c %s 2>&1 | FileCheck -check-prefix=GENERICV8A-NOSVE %s
 // GENERICV8A-NOSVE-NOT: "-target-feature" "+sve"
 
-// The 32-bit floating point matrix multiply extension is enabled by default
-// for armv8.6-a targets (or later) with SVE, and can optionally be enabled for
-// any target from armv8.2a onwards (we don't enforce not using it with earlier
-// targets).
+// The 32-bit floating point matrix multiply extension is an optional feature
+// that can be used for any target from armv8.2a and onwards. This can be
+// enabled using the `+f32mm` option.`.
 // RUN: %clang --target=aarch64 -march=armv8.6a       -### -c %s 2>&1 | FileCheck -check-prefix=NO-F32MM %s
-// RUN: %clang --target=aarch64 -march=armv8.6a+sve   -### -c %s 2>&1 | FileCheck -check-prefix=F32MM %s
+// RUN: %clang --target=aarch64 -march=armv8.6a+sve+f32mm   -### -c %s 2>&1 | FileCheck -check-prefix=F32MM %s
 // RUN: %clang --target=aarch64 -march=armv8.5a+f32mm -### -c %s 2>&1 | FileCheck -check-prefix=F32MM %s
 // NO-F32MM-NOT: "-target-feature" "+f32mm"
 // F32MM: "-target-feature" "+f32mm"

diff --git a/clang/test/InstallAPI/Inputs/Umbrella/Umbrella.framework/Headers/AAA.h b/clang/test/InstallAPI/Inputs/Umbrella/Umbrella.framework/Headers/AAA.h
@@ -0,0 +1,3 @@
+#ifndef PUBLIC_UMBRELLA_HEADER_FIRST
+#error "Public umbrella header was not included first!"
+#endif
diff --git a/clang/test/InstallAPI/Inputs/Umbrella/Umbrella.framework/Headers/SpecialUmbrella.h b/clang/test/InstallAPI/Inputs/Umbrella/Umbrella.framework/Headers/SpecialUmbrella.h
@@ -0,0 +1 @@
+#define PUBLIC_UMBRELLA_HEADER_FIRST
diff --git a/clang/test/InstallAPI/Inputs/Umbrella/Umbrella.framework/PrivateHeaders/AAA_Private.h b/clang/test/InstallAPI/Inputs/Umbrella/Umbrella.framework/PrivateHeaders/AAA_Private.h
@@ -0,0 +1,3 @@
+#ifndef PRIVATE_UMBRELLA_HEADER_FIRST
+#error "Private umbrella header was not included first!"
+#endif
diff --git a/...est/InstallAPI/Inputs/Umbrella/Umbrella.framework/PrivateHeaders/SpecialPrivateUmbrella.h b/...est/InstallAPI/Inputs/Umbrella/Umbrella.framework/PrivateHeaders/SpecialPrivateUmbrella.h
@@ -0,0 +1 @@
+#define PRIVATE_UMBRELLA_HEADER_FIRST
diff --git a/clang/test/InstallAPI/umbrella-headers-unix.test b/clang/test/InstallAPI/umbrella-headers-unix.test
@@ -0,0 +1,40 @@
+// UNSUPPORTED: system-windows
+
+; RUN: rm -rf %t
+; RUN: split-file %s %t
+; RUN: sed -e "s|DSTROOT|%/t|g" %t/inputs.json.in > %t/inputs.json
+; RUN: mkdir %t/Frameworks/
+; RUN: cp -r %S/Inputs/Umbrella/Umbrella.framework %t/Frameworks/
+
+// Only validate path based input that rely on regex matching on unix based file systems.
+; RUN: clang-installapi --target=arm64-apple-macosx13 \
+; RUN:  -install_name /System/Library/Frameworks/Umbrella2.framework/Versions/A/Umbrella \
+; RUN: -ObjC -F%t/Frameworks/ %t/inputs.json \
+; RUN: --public-umbrella-header=%t/Frameworks/Umbrella.framework/Headers/SpecialUmbrella.h \
+; RUN: -private-umbrella-header \
+; RUN: %t/Frameworks/Umbrella.framework/PrivateHeaders/SpecialPrivateUmbrella.h \
+; RUN: -o %t/output.tbd 2>&1 | FileCheck -allow-empty %s
+
+; CHECK-NOT: error
+; CHECK-NOT: warning
+
+;--- inputs.json.in
+{
+  "headers": [ {
+    "path" : "DSTROOT/Frameworks/Umbrella.framework/Headers/AAA.h",
+    "type" : "public"
+  }, 
+  {
+    "path" : "DSTROOT/Frameworks/Umbrella.framework/Headers/SpecialUmbrella.h",
+    "type" : "public"
+  },
+  {
+    "path" : "DSTROOT/Frameworks/Umbrella.framework/PrivateHeaders/AAA_Private.h",
+    "type" : "private"
+  },
+  {
+    "path" : "DSTROOT/Frameworks/Umbrella.framework/PrivateHeaders/SpecialPrivateUmbrella.h",
+    "type" : "private"
+  }],
+  "version": "3"
+}
diff --git a/clang/test/InstallAPI/umbrella-headers.test b/clang/test/InstallAPI/umbrella-headers.test
@@ -0,0 +1,48 @@
+; RUN: rm -rf %t
+; RUN: split-file %s %t
+; RUN: sed -e "s|DSTROOT|%/t|g" %t/inputs.json.in > %t/inputs.json
+; RUN: cp -r %S/Inputs/Umbrella/Umbrella.framework %t/Frameworks/
+
+// Check base filename matches.
+; RUN: clang-installapi --target=arm64-apple-macosx13 \
+; RUN: -install_name /System/Library/Frameworks/Umbrella.framework/Versions/A/Umbrella \
+; RUN: -ObjC -F%t/Frameworks/ %t/inputs.json \
+; RUN: --public-umbrella-header=SpecialUmbrella.h \
+; RUN: --private-umbrella-header=SpecialPrivateUmbrella.h \
+; RUN: -o %t/output.tbd 2>&1 | FileCheck -allow-empty %s
+
+// Try missing umbrella header argument.
+; RUN: not clang-installapi --target=arm64-apple-macosx13 \
+; RUN: -install_name /System/Library/Frameworks/Umbrella.framework/Versions/A/Umbrella \
+; RUN: -ObjC -F%t/Frameworks/ %t/inputs.json \
+; RUN: --public-umbrella-header=Ignore.h \
+; RUN: -o %t/output.tbd 2>&1 | FileCheck %s -check-prefix=ERR
+
+; ERR: error: public umbrella header file not found in input: 'Ignore.h'
+
+; CHECK-NOT: error
+; CHECK-NOT: warning
+
+;--- Frameworks/Umbrella.framework/Headers/Ignore.h
+#error "This header should be ignored"
+
+;--- inputs.json.in
+{
+  "headers": [ {
+    "path" : "DSTROOT/Frameworks/Umbrella.framework/Headers/AAA.h",
+    "type" : "public"
+  }, 
+  {
+    "path" : "DSTROOT/Frameworks/Umbrella.framework/Headers/SpecialUmbrella.h",
+    "type" : "public"
+  },
+  {
+    "path" : "DSTROOT/Frameworks/Umbrella.framework/PrivateHeaders/AAA_Private.h",
+    "type" : "private"
+  },
+  {
+    "path" : "DSTROOT/Frameworks/Umbrella.framework/PrivateHeaders/SpecialPrivateUmbrella.h",
+    "type" : "private"
+  }],
+  "version": "3"
+}
diff --git a/clang/test/Modules/codegen.test b/clang/test/Modules/codegen.test
@@ -26,7 +26,7 @@ USE: $_Z4instIiEvv = comdat any
 USE: $_Z10always_inlv = comdat any
 FOO: $_ZN13implicit_dtorD2Ev = comdat any
 FOO: define weak_odr void @_Z2f1PKcz(ptr noundef %fmt, ...) #{{[0-9]+}} comdat
-FOO:   call void @llvm.va_start(ptr %{{[a-zA-Z0-9]*}})
+FOO:   call void @llvm.va_start.p0(ptr %{{[a-zA-Z0-9]*}})
 
 Test that implicit special members are emitted into the FOO module if they're
 ODR used there, otherwise emit them linkonce_odr as usual in the use.

diff --git a/clang/test/Preprocessor/aarch64-target-features.c b/clang/test/Preprocessor/aarch64-target-features.c
@@ -196,7 +196,7 @@
 // CHECK-8_6-NOT: __ARM_FEATURE_SHA3 1
 // CHECK-8_6-NOT: __ARM_FEATURE_SM4 1
 
-// RUN: %clang -target aarch64-none-linux-gnu -march=armv8.6-a+sve -x c -E -dM %s -o - | FileCheck --check-prefix=CHECK-SVE-8_6 %s
+// RUN: %clang -target aarch64-none-linux-gnu -march=armv8.6-a+sve+f32mm -x c -E -dM %s -o - | FileCheck --check-prefix=CHECK-SVE-8_6 %s
 // CHECK-SVE-8_6: __ARM_FEATURE_SVE 1
 // CHECK-SVE-8_6: __ARM_FEATURE_SVE_BF16 1
 // CHECK-SVE-8_6: __ARM_FEATURE_SVE_MATMUL_FP32 1

diff --git a/clang/test/SemaTemplate/concepts.cpp b/clang/test/SemaTemplate/concepts.cpp
@@ -1,4 +1,4 @@
-// RUN: %clang_cc1 -std=c++20 -verify %s
+// RUN: %clang_cc1 -std=c++20 -ferror-limit 0 -verify %s
 
 namespace PR47043 {
   template<typename T> concept True = true;
@@ -1114,3 +1114,11 @@ void foo() {
 }
 
 } // namespace GH64808
+
+namespace GH86757_1 {
+template <typename...> concept b = false;
+template <typename> concept c = b<>;
+template <typename d> concept f = c< d >;
+template <f> struct e; // expected-note {{}}
+template <f d> struct e<d>; // expected-error {{class template partial specialization is not more specialized than the primary template}}
+}
diff --git a/clang/tools/clang-installapi/InstallAPIOpts.td b/clang/tools/clang-installapi/InstallAPIOpts.td
@@ -61,3 +61,15 @@ def exclude_private_header : Separate<["-"], "exclude-private-header">,
   HelpText<"Exclude private header from parsing">;
 def exclude_private_header_EQ : Joined<["--"], "exclude-private-header=">,
   Alias<exclude_private_header>;
+def public_umbrella_header : Separate<["-"], "public-umbrella-header">,
+  MetaVarName<"<path>">, HelpText<"Specify the public umbrella header location">;
+def public_umbrella_header_EQ : Joined<["--"], "public-umbrella-header=">,
+  Alias<public_umbrella_header>;
+def private_umbrella_header : Separate<["-"], "private-umbrella-header">,
+  MetaVarName<"<path>">, HelpText<"Specify the private umbrella header location">;
+def private_umbrella_header_EQ : Joined<["--"], "private-umbrella-header=">,
+  Alias<private_umbrella_header>;
+def project_umbrella_header : Separate<["-"], "project-umbrella-header">,
+  MetaVarName<"<path>">, HelpText<"Specify the project umbrella header location">;
+def project_umbrella_header_EQ : Joined<["--"], "project-umbrella-header=">,
+  Alias<project_umbrella_header>;
diff --git a/clang/tools/clang-installapi/Options.cpp b/clang/tools/clang-installapi/Options.cpp
@@ -270,6 +270,16 @@ Options::processAndFilterOutInstallAPIOptions(ArrayRef<const char *> Args) {
                                  OPT_exclude_project_header))
     return {};
 
+  // Handle umbrella headers.
+  if (const Arg *A = ParsedArgs.getLastArg(OPT_public_umbrella_header))
+    DriverOpts.PublicUmbrellaHeader = A->getValue();
+
+  if (const Arg *A = ParsedArgs.getLastArg(OPT_private_umbrella_header))
+    DriverOpts.PrivateUmbrellaHeader = A->getValue();
+
+  if (const Arg *A = ParsedArgs.getLastArg(OPT_project_umbrella_header))
+    DriverOpts.ProjectUmbrellaHeader = A->getValue();
+
   /// Any unclaimed arguments should be forwarded to the clang driver.
   std::vector<const char *> ClangDriverArgs(ParsedArgs.size());
   for (const Arg *A : ParsedArgs) {
@@ -323,6 +333,15 @@ Options::Options(DiagnosticsEngine &Diag, FileManager *FM,
   }
 }
 
+static const Regex Rule("(.+)/(.+)\\.framework/");
+static StringRef getFrameworkNameFromInstallName(StringRef InstallName) {
+  SmallVector<StringRef, 3> Match;
+  Rule.match(InstallName, &Match);
+  if (Match.empty())
+    return "";
+  return Match.back();
+}
+
 InstallAPIContext Options::createContext() {
   InstallAPIContext Ctx;
   Ctx.FM = FM;
@@ -339,6 +358,11 @@ InstallAPIContext Options::createContext() {
   Ctx.OutputLoc = DriverOpts.OutputPath;
   Ctx.LangMode = FEOpts.LangMode;
 
+  // Attempt to find umbrella headers by capturing framework name.
+  StringRef FrameworkName;
+  if (!LinkerOpts.IsDylib)
+    FrameworkName = getFrameworkNameFromInstallName(LinkerOpts.InstallName);
+
   // Process inputs.
   for (const std::string &ListPath : DriverOpts.FileLists) {
     auto Buffer = FM->getBufferForFile(ListPath);
@@ -357,8 +381,7 @@ InstallAPIContext Options::createContext() {
     assert(Type != HeaderType::Unknown && "Missing header type.");
     for (const StringRef Path : Headers) {
       if (!FM->getOptionalFileRef(Path)) {
-        Diags->Report(diag::err_no_such_header_file)
-            << Path << (unsigned)Type - 1;
+        Diags->Report(diag::err_no_such_header_file) << Path << (unsigned)Type;
         return false;
       }
       SmallString<PATH_MAX> FullPath(Path);
@@ -382,6 +405,7 @@ InstallAPIContext Options::createContext() {
   std::vector<std::unique_ptr<HeaderGlob>> ExcludedHeaderGlobs;
   std::set<FileEntryRef> ExcludedHeaderFiles;
   auto ParseGlobs = [&](const PathSeq &Paths, HeaderType Type) {
+    assert(Type != HeaderType::Unknown && "Missing header type.");
     for (const StringRef Path : Paths) {
       auto Glob = HeaderGlob::create(Path, Type);
       if (Glob)
@@ -424,6 +448,57 @@ InstallAPIContext Options::createContext() {
     if (!Glob->didMatch())
       Diags->Report(diag::warn_glob_did_not_match) << Glob->str();
 
+  // Mark any explicit or inferred umbrella headers. If one exists, move
+  // that to the beginning of the input headers.
+  auto MarkandMoveUmbrellaInHeaders = [&](llvm::Regex &Regex,
+                                          HeaderType Type) -> bool {
+    auto It = find_if(Ctx.InputHeaders, [&Regex, Type](const HeaderFile &H) {
+      return (H.getType() == Type) && Regex.match(H.getPath());
+    });
+
+    if (It == Ctx.InputHeaders.end())
+      return false;
+    It->setUmbrellaHeader();
+
+    // Because there can be an umbrella header per header type,
+    // find the first non umbrella header to swap position with.
+    auto BeginPos = find_if(Ctx.InputHeaders, [](const HeaderFile &H) {
+      return !H.isUmbrellaHeader();
+    });
+    if (BeginPos != Ctx.InputHeaders.end() && BeginPos < It)
+      std::swap(*BeginPos, *It);
+    return true;
+  };
+
+  auto FindUmbrellaHeader = [&](StringRef HeaderPath, HeaderType Type) -> bool {
+    assert(Type != HeaderType::Unknown && "Missing header type.");
+    if (!HeaderPath.empty()) {
+      auto EscapedString = Regex::escape(HeaderPath);
+      Regex UmbrellaRegex(EscapedString);
+      if (!MarkandMoveUmbrellaInHeaders(UmbrellaRegex, Type)) {
+        Diags->Report(diag::err_no_such_umbrella_header_file)
+            << HeaderPath << (unsigned)Type;
+        return false;
+      }
+    } else if (!FrameworkName.empty() && (Type != HeaderType::Project)) {
+      auto UmbrellaName = "/" + Regex::escape(FrameworkName);
+      if (Type == HeaderType::Public)
+        UmbrellaName += "\\.h";
+      else
+        UmbrellaName += "[_]?Private\\.h";
+      Regex UmbrellaRegex(UmbrellaName);
+      MarkandMoveUmbrellaInHeaders(UmbrellaRegex, Type);
+    }
+    return true;
+  };
+  if (!FindUmbrellaHeader(DriverOpts.PublicUmbrellaHeader,
+                          HeaderType::Public) ||
+      !FindUmbrellaHeader(DriverOpts.PrivateUmbrellaHeader,
+                          HeaderType::Private) ||
+      !FindUmbrellaHeader(DriverOpts.ProjectUmbrellaHeader,
+                          HeaderType::Project))
+    return Ctx;
+
   // Parse binary dylib and initialize verifier.
   if (DriverOpts.DylibToVerify.empty()) {
     Ctx.Verifier = std::make_unique<DylibVerifier>();

diff --git a/clang/tools/clang-installapi/Options.h b/clang/tools/clang-installapi/Options.h
@@ -31,6 +31,15 @@ struct DriverOptions {
   /// \brief Path to input file lists (JSON).
   llvm::MachO::PathSeq FileLists;
 
+  /// \brief Path to public umbrella header.
+  std::string PublicUmbrellaHeader;
+
+  /// \brief Path to private umbrella header.
+  std::string PrivateUmbrellaHeader;
+
+  /// \brief Path to project umbrella header.
+  std::string ProjectUmbrellaHeader;
+
   /// \brief Paths of extra public headers.
   PathSeq ExtraPublicHeaders;
 

diff --git a/clang/tools/libclang/CXType.cpp b/clang/tools/libclang/CXType.cpp
@@ -680,6 +680,7 @@ CXCallingConv clang_getFunctionTypeCallingConv(CXType X) {
       TCALLINGCONV(PreserveAll);
       TCALLINGCONV(M68kRTD);
       TCALLINGCONV(PreserveNone);
+      TCALLINGCONV(RISCVVectorCall);
     case CC_SpirFunction: return CXCallingConv_Unexposed;
     case CC_AMDGPUKernelCall: return CXCallingConv_Unexposed;
     case CC_OpenCLKernel: return CXCallingConv_Unexposed;

diff --git a/clang/utils/TableGen/RISCVVEmitter.cpp b/clang/utils/TableGen/RISCVVEmitter.cpp
@@ -334,10 +334,6 @@ void RVVEmitter::createHeader(raw_ostream &OS) {
   OS << "#include <stdint.h>\n";
   OS << "#include <stddef.h>\n\n";
 
-  OS << "#ifndef __riscv_vector\n";
-  OS << "#error \"Vector intrinsics require the vector extension.\"\n";
-  OS << "#endif\n\n";
-
   OS << "#ifdef __cplusplus\n";
   OS << "extern \"C\" {\n";
   OS << "#endif\n\n";

diff --git a/compiler-rt/lib/scudo/standalone/tests/strings_test.cpp b/compiler-rt/lib/scudo/standalone/tests/strings_test.cpp
@@ -141,7 +141,7 @@ TEST(ScudoStringsTest, CapacityIncreaseFails) {
 
   // Test requires that the default length is at least 6 characters.
   scudo::uptr MaxSize = Str.capacity();
-  EXPECT_LE(6, MaxSize);
+  EXPECT_LE(6u, MaxSize);
 
   for (size_t i = 0; i < MaxSize - 5; i++) {
     Str.append("B");

diff --git a/flang/lib/Lower/OpenMP/ClauseProcessor.h b/flang/lib/Lower/OpenMP/ClauseProcessor.h
@@ -51,7 +51,7 @@ class ClauseProcessor {
                   Fortran::semantics::SemanticsContext &semaCtx,
                   const Fortran::parser::OmpClauseList &clauses)
       : converter(converter), semaCtx(semaCtx),
-        clauses(makeList(clauses, semaCtx)) {}
+        clauses(makeClauses(clauses, semaCtx)) {}
 
   // 'Unique' clauses: They can appear at most once in the clause list.
   bool processCollapse(

diff --git a/flang/lib/Lower/OpenMP/Clauses.cpp b/flang/lib/Lower/OpenMP/Clauses.cpp
@@ -347,7 +347,7 @@ Aligned make(const parser::OmpClause::Aligned &inp,
 
   return Aligned{{
       /*Alignment=*/maybeApply(makeExprFn(semaCtx), t1),
-      /*List=*/makeList(t0, semaCtx),
+      /*List=*/makeObjects(t0, semaCtx),
   }};
 }
 
@@ -362,7 +362,7 @@ Allocate make(const parser::OmpClause::Allocate &inp,
     return Allocate{{/*AllocatorSimpleModifier=*/std::nullopt,
                      /*AllocatorComplexModifier=*/std::nullopt,
                      /*AlignModifier=*/std::nullopt,
-                     /*List=*/makeList(t1, semaCtx)}};
+                     /*List=*/makeObjects(t1, semaCtx)}};
   }
 
   using Tuple = decltype(Allocate::t);
@@ -374,7 +374,7 @@ Allocate make(const parser::OmpClause::Allocate &inp,
             return {/*AllocatorSimpleModifier=*/makeExpr(v.v, semaCtx),
                     /*AllocatorComplexModifier=*/std::nullopt,
                     /*AlignModifier=*/std::nullopt,
-                    /*List=*/makeList(t1, semaCtx)};
+                    /*List=*/makeObjects(t1, semaCtx)};
           },
           // complex-modifier + align-modifier
           [&](const wrapped::AllocateModifier::ComplexModifier &v) -> Tuple {
@@ -384,14 +384,14 @@ Allocate make(const parser::OmpClause::Allocate &inp,
                 /*AllocatorSimpleModifier=*/std::nullopt,
                 /*AllocatorComplexModifier=*/Allocator{makeExpr(s0.v, semaCtx)},
                 /*AlignModifier=*/Align{makeExpr(s1.v, semaCtx)},
-                /*List=*/makeList(t1, semaCtx)};
+                /*List=*/makeObjects(t1, semaCtx)};
           },
           // align-modifier
           [&](const wrapped::AllocateModifier::Align &v) -> Tuple {
             return {/*AllocatorSimpleModifier=*/std::nullopt,
                     /*AllocatorComplexModifier=*/std::nullopt,
                     /*AlignModifier=*/Align{makeExpr(v.v, semaCtx)},
-                    /*List=*/makeList(t1, semaCtx)};
+                    /*List=*/makeObjects(t1, semaCtx)};
           },
       },
       t0->u)};
@@ -450,13 +450,13 @@ Collapse make(const parser::OmpClause::Collapse &inp,
 Copyin make(const parser::OmpClause::Copyin &inp,
             semantics::SemanticsContext &semaCtx) {
   // inp.v -> parser::OmpObjectList
-  return Copyin{/*List=*/makeList(inp.v, semaCtx)};
+  return Copyin{/*List=*/makeObjects(inp.v, semaCtx)};
 }
 
 Copyprivate make(const parser::OmpClause::Copyprivate &inp,
                  semantics::SemanticsContext &semaCtx) {
   // inp.v -> parser::OmpObjectList
-  return Copyprivate{/*List=*/makeList(inp.v, semaCtx)};
+  return Copyprivate{/*List=*/makeObjects(inp.v, semaCtx)};
 }
 
 Default make(const parser::OmpClause::Default &inp,
@@ -641,7 +641,7 @@ Doacross make(const parser::OmpClause::Doacross &inp,
 Enter make(const parser::OmpClause::Enter &inp,
            semantics::SemanticsContext &semaCtx) {
   // inp.v -> parser::OmpObjectList
-  return Enter{makeList(/*List=*/inp.v, semaCtx)};
+  return Enter{makeObjects(/*List=*/inp.v, semaCtx)};
 }
 
 Exclusive make(const parser::OmpClause::Exclusive &inp,
@@ -671,7 +671,7 @@ Final make(const parser::OmpClause::Final &inp,
 Firstprivate make(const parser::OmpClause::Firstprivate &inp,
                   semantics::SemanticsContext &semaCtx) {
   // inp.v -> parser::OmpObjectList
-  return Firstprivate{/*List=*/makeList(inp.v, semaCtx)};
+  return Firstprivate{/*List=*/makeObjects(inp.v, semaCtx)};
 }
 
 // Flush: empty
@@ -681,7 +681,7 @@ From make(const parser::OmpClause::From &inp,
   // inp.v -> parser::OmpObjectList
   return From{{/*Expectation=*/std::nullopt, /*Mapper=*/std::nullopt,
                /*Iterator=*/std::nullopt,
-               /*LocatorList=*/makeList(inp.v, semaCtx)}};
+               /*LocatorList=*/makeObjects(inp.v, semaCtx)}};
 }
 
 // Full: empty
@@ -696,7 +696,7 @@ Grainsize make(const parser::OmpClause::Grainsize &inp,
 HasDeviceAddr make(const parser::OmpClause::HasDeviceAddr &inp,
                    semantics::SemanticsContext &semaCtx) {
   // inp.v -> parser::OmpObjectList
-  return HasDeviceAddr{/*List=*/makeList(inp.v, semaCtx)};
+  return HasDeviceAddr{/*List=*/makeObjects(inp.v, semaCtx)};
 }
 
 Hint make(const parser::OmpClause::Hint &inp,
@@ -762,20 +762,20 @@ InReduction make(const parser::OmpClause::InReduction &inp,
   auto &t1 = std::get<parser::OmpObjectList>(inp.v.t);
   return InReduction{
       {/*ReductionIdentifiers=*/{makeReductionOperator(t0, semaCtx)},
-       /*List=*/makeList(t1, semaCtx)}};
+       /*List=*/makeObjects(t1, semaCtx)}};
 }
 
 IsDevicePtr make(const parser::OmpClause::IsDevicePtr &inp,
                  semantics::SemanticsContext &semaCtx) {
   // inp.v -> parser::OmpObjectList
-  return IsDevicePtr{/*List=*/makeList(inp.v, semaCtx)};
+  return IsDevicePtr{/*List=*/makeObjects(inp.v, semaCtx)};
 }
 
 Lastprivate make(const parser::OmpClause::Lastprivate &inp,
                  semantics::SemanticsContext &semaCtx) {
   // inp.v -> parser::OmpObjectList
   return Lastprivate{{/*LastprivateModifier=*/std::nullopt,
-                      /*List=*/makeList(inp.v, semaCtx)}};
+                      /*List=*/makeObjects(inp.v, semaCtx)}};
 }
 
 Linear make(const parser::OmpClause::Linear &inp,
@@ -817,7 +817,7 @@ Linear make(const parser::OmpClause::Linear &inp,
 Link make(const parser::OmpClause::Link &inp,
           semantics::SemanticsContext &semaCtx) {
   // inp.v -> parser::OmpObjectList
-  return Link{/*List=*/makeList(inp.v, semaCtx)};
+  return Link{/*List=*/makeObjects(inp.v, semaCtx)};
 }
 
 Map make(const parser::OmpClause::Map &inp,
@@ -844,7 +844,7 @@ Map make(const parser::OmpClause::Map &inp,
   if (!t0) {
     return Map{{/*MapType=*/std::nullopt, /*MapTypeModifiers=*/std::nullopt,
                 /*Mapper=*/std::nullopt, /*Iterator=*/std::nullopt,
-                /*LocatorList=*/makeList(t1, semaCtx)}};
+                /*LocatorList=*/makeObjects(t1, semaCtx)}};
   }
 
   auto &s0 = std::get<std::optional<parser::OmpMapType::Always>>(t0->t);
@@ -857,7 +857,7 @@ Map make(const parser::OmpClause::Map &inp,
   return Map{{/*MapType=*/convert1(s1),
               /*MapTypeModifiers=*/maybeList,
               /*Mapper=*/std::nullopt, /*Iterator=*/std::nullopt,
-              /*LocatorList=*/makeList(t1, semaCtx)}};
+              /*LocatorList=*/makeObjects(t1, semaCtx)}};
 }
 
 // Match: incomplete
@@ -980,7 +980,7 @@ Priority make(const parser::OmpClause::Priority &inp,
 Private make(const parser::OmpClause::Private &inp,
              semantics::SemanticsContext &semaCtx) {
   // inp.v -> parser::OmpObjectList
-  return Private{/*List=*/makeList(inp.v, semaCtx)};
+  return Private{/*List=*/makeObjects(inp.v, semaCtx)};
 }
 
 ProcBind make(const parser::OmpClause::ProcBind &inp,
@@ -1010,7 +1010,7 @@ Reduction make(const parser::OmpClause::Reduction &inp,
   return Reduction{
       {/*ReductionIdentifiers=*/{makeReductionOperator(t0, semaCtx)},
        /*ReductionModifier=*/std::nullopt,
-       /*List=*/makeList(t1, semaCtx)}};
+       /*List=*/makeObjects(t1, semaCtx)}};
 }
 
 // Relaxed: empty
@@ -1104,7 +1104,7 @@ Severity make(const parser::OmpClause::Severity &inp,
 Shared make(const parser::OmpClause::Shared &inp,
             semantics::SemanticsContext &semaCtx) {
   // inp.v -> parser::OmpObjectList
-  return Shared{/*List=*/makeList(inp.v, semaCtx)};
+  return Shared{/*List=*/makeObjects(inp.v, semaCtx)};
 }
 
 // Simd: empty
@@ -1128,7 +1128,7 @@ TaskReduction make(const parser::OmpClause::TaskReduction &inp,
   auto &t1 = std::get<parser::OmpObjectList>(inp.v.t);
   return TaskReduction{
       {/*ReductionIdentifiers=*/{makeReductionOperator(t0, semaCtx)},
-       /*List=*/makeList(t1, semaCtx)}};
+       /*List=*/makeObjects(t1, semaCtx)}};
 }
 
 ThreadLimit make(const parser::OmpClause::ThreadLimit &inp,
@@ -1145,7 +1145,7 @@ To make(const parser::OmpClause::To &inp,
   // inp.v -> parser::OmpObjectList
   return To{{/*Expectation=*/std::nullopt, /*Mapper=*/std::nullopt,
              /*Iterator=*/std::nullopt,
-             /*LocatorList=*/makeList(inp.v, semaCtx)}};
+             /*LocatorList=*/makeObjects(inp.v, semaCtx)}};
 }
 
 // UnifiedAddress: empty
@@ -1175,13 +1175,13 @@ Use make(const parser::OmpClause::Use &inp,
 UseDeviceAddr make(const parser::OmpClause::UseDeviceAddr &inp,
                    semantics::SemanticsContext &semaCtx) {
   // inp.v -> parser::OmpObjectList
-  return UseDeviceAddr{/*List=*/makeList(inp.v, semaCtx)};
+  return UseDeviceAddr{/*List=*/makeObjects(inp.v, semaCtx)};
 }
 
 UseDevicePtr make(const parser::OmpClause::UseDevicePtr &inp,
                   semantics::SemanticsContext &semaCtx) {
   // inp.v -> parser::OmpObjectList
-  return UseDevicePtr{/*List=*/makeList(inp.v, semaCtx)};
+  return UseDevicePtr{/*List=*/makeObjects(inp.v, semaCtx)};
 }
 
 UsesAllocators make(const parser::OmpClause::UsesAllocators &inp,
@@ -1205,8 +1205,8 @@ Clause makeClause(const Fortran::parser::OmpClause &cls,
       cls.u);
 }
 
-List<Clause> makeList(const parser::OmpClauseList &clauses,
-                      semantics::SemanticsContext &semaCtx) {
+List<Clause> makeClauses(const parser::OmpClauseList &clauses,
+                         semantics::SemanticsContext &semaCtx) {
   return makeList(clauses.v, [&](const parser::OmpClause &s) {
     return makeClause(s, semaCtx);
   });

diff --git a/flang/lib/Lower/OpenMP/Clauses.h b/flang/lib/Lower/OpenMP/Clauses.h
@@ -88,8 +88,8 @@ List<ResultTy> makeList(ContainerTy &&container, FunctionTy &&func) {
   return v;
 }
 
-inline ObjectList makeList(const parser::OmpObjectList &objects,
-                           semantics::SemanticsContext &semaCtx) {
+inline ObjectList makeObjects(const parser::OmpObjectList &objects,
+                              semantics::SemanticsContext &semaCtx) {
   return makeList(objects.v, makeObjectFn(semaCtx));
 }
 
@@ -256,8 +256,8 @@ Clause makeClause(llvm::omp::Clause id, Specific &&specific,
 Clause makeClause(const Fortran::parser::OmpClause &cls,
                   semantics::SemanticsContext &semaCtx);
 
-List<Clause> makeList(const parser::OmpClauseList &clauses,
-                      semantics::SemanticsContext &semaCtx);
+List<Clause> makeClauses(const parser::OmpClauseList &clauses,
+                         semantics::SemanticsContext &semaCtx);
 } // namespace Fortran::lower::omp
 
 #endif // FORTRAN_LOWER_OPENMP_CLAUSES_H
diff --git a/flang/lib/Lower/OpenMP/DataSharingProcessor.h b/flang/lib/Lower/OpenMP/DataSharingProcessor.h
@@ -89,7 +89,7 @@ class DataSharingProcessor {
                        Fortran::lower::SymMap *symTable = nullptr)
       : hasLastPrivateOp(false), converter(converter),
         firOpBuilder(converter.getFirOpBuilder()),
-        clauses(omp::makeList(opClauseList, semaCtx)), eval(eval),
+        clauses(omp::makeClauses(opClauseList, semaCtx)), eval(eval),
         useDelayedPrivatization(useDelayedPrivatization), symTable(symTable) {}
 
   // Privatisation is split into two steps.

diff --git a/flang/lib/Lower/OpenMP/OpenMP.cpp b/flang/lib/Lower/OpenMP/OpenMP.cpp
@@ -1254,7 +1254,7 @@ static mlir::omp::DeclareTargetDeviceType getDeclareTargetInfo(
 
   if (const auto *objectList{
           Fortran::parser::Unwrap<Fortran::parser::OmpObjectList>(spec.u)}) {
-    ObjectList objects{makeList(*objectList, semaCtx)};
+    ObjectList objects{makeObjects(*objectList, semaCtx)};
     // Case: declare target(func, var1, var2)
     gatherFuncAndVarSyms(objects, mlir::omp::DeclareTargetCaptureClause::to,
                          symbolAndClause);
@@ -2352,7 +2352,7 @@ void Fortran::lower::genOpenMPReduction(
     const Fortran::parser::OmpClauseList &clauseList) {
   fir::FirOpBuilder &firOpBuilder = converter.getFirOpBuilder();
 
-  List<Clause> clauses{makeList(clauseList, semaCtx)};
+  List<Clause> clauses{makeClauses(clauseList, semaCtx)};
 
   for (const Clause &clause : clauses) {
     if (const auto &reductionClause =

diff --git a/libc/docs/dev/code_style.rst b/libc/docs/dev/code_style.rst
@@ -55,7 +55,7 @@ We define two kinds of macros:
    * ``src/__support/macros/config.h`` - Important compiler and platform
      features. Such macros can be used to produce portable code by
      parameterizing compilation based on the presence or lack of a given
-     feature. e.g., ``LIBC_HAS_BUILTIN``
+     feature. e.g., ``LIBC_HAS_FEATURE``
    * ``src/__support/macros/attributes.h`` - Attributes for functions, types,
      and variables. e.g., ``LIBC_UNUSED``
    * ``src/__support/macros/optimization.h`` - Portable macros for performance

diff --git a/libc/src/__support/CPP/CMakeLists.txt b/libc/src/__support/CPP/CMakeLists.txt
@@ -18,7 +18,6 @@ add_header_library(
     .limits
     .type_traits
     libc.src.__support.macros.attributes
-    libc.src.__support.macros.config
     libc.src.__support.macros.sanitizer
 )
 
@@ -157,7 +156,6 @@ add_header_library(
   DEPENDS
     libc.include.llvm-libc-macros.stdfix_macros
     libc.src.__support.macros.attributes
-    libc.src.__support.macros.config
     libc.src.__support.macros.properties.types
 )
 

diff --git a/libc/src/__support/CPP/atomic.h b/libc/src/__support/CPP/atomic.h
@@ -71,10 +71,11 @@ template <typename T> struct Atomic {
 
   T load(MemoryOrder mem_ord = MemoryOrder::SEQ_CST,
          [[maybe_unused]] MemoryScope mem_scope = MemoryScope::DEVICE) {
-    if constexpr (LIBC_HAS_BUILTIN(__scoped_atomic_load_n))
-      return __scoped_atomic_load_n(&val, int(mem_ord), (int)(mem_scope));
-    else
-      return __atomic_load_n(&val, int(mem_ord));
+#if __has_builtin(__scoped_atomic_load_n)
+    return __scoped_atomic_load_n(&val, int(mem_ord), (int)(mem_scope));
+#else
+    return __atomic_load_n(&val, int(mem_ord));
+#endif
   }
 
   // Atomic store.
@@ -85,10 +86,11 @@ template <typename T> struct Atomic {
 
   void store(T rhs, MemoryOrder mem_ord = MemoryOrder::SEQ_CST,
              [[maybe_unused]] MemoryScope mem_scope = MemoryScope::DEVICE) {
-    if constexpr (LIBC_HAS_BUILTIN(__scoped_atomic_store_n))
-      __scoped_atomic_store_n(&val, rhs, int(mem_ord), (int)(mem_scope));
-    else
-      __atomic_store_n(&val, rhs, int(mem_ord));
+#if __has_builtin(__scoped_atomic_store_n)
+    __scoped_atomic_store_n(&val, rhs, int(mem_ord), (int)(mem_scope));
+#else
+    __atomic_store_n(&val, rhs, int(mem_ord));
+#endif
   }
 
   // Atomic compare exchange
@@ -101,47 +103,51 @@ template <typename T> struct Atomic {
 
   T exchange(T desired, MemoryOrder mem_ord = MemoryOrder::SEQ_CST,
              [[maybe_unused]] MemoryScope mem_scope = MemoryScope::DEVICE) {
-    if constexpr (LIBC_HAS_BUILTIN(__scoped_atomic_exchange_n))
-      return __scoped_atomic_exchange_n(&val, desired, int(mem_ord),
-                                        (int)(mem_scope));
-    else
-      return __atomic_exchange_n(&val, desired, int(mem_ord));
+#if __has_builtin(__scoped_atomic_exchange_n)
+    return __scoped_atomic_exchange_n(&val, desired, int(mem_ord),
+                                      (int)(mem_scope));
+#else
+    return __atomic_exchange_n(&val, desired, int(mem_ord));
+#endif
   }
 
   T fetch_add(T increment, MemoryOrder mem_ord = MemoryOrder::SEQ_CST,
               [[maybe_unused]] MemoryScope mem_scope = MemoryScope::DEVICE) {
-    if constexpr (LIBC_HAS_BUILTIN(__scoped_atomic_fetch_add))
-      return __scoped_atomic_fetch_add(&val, increment, int(mem_ord),
-                                       (int)(mem_scope));
-    else
-      return __atomic_fetch_add(&val, increment, int(mem_ord));
+#if __has_builtin(__scoped_atomic_fetch_add)
+    return __scoped_atomic_fetch_add(&val, increment, int(mem_ord),
+                                     (int)(mem_scope));
+#else
+    return __atomic_fetch_add(&val, increment, int(mem_ord));
+#endif
   }
 
   T fetch_or(T mask, MemoryOrder mem_ord = MemoryOrder::SEQ_CST,
              [[maybe_unused]] MemoryScope mem_scope = MemoryScope::DEVICE) {
-    if constexpr (LIBC_HAS_BUILTIN(__scoped_atomic_fetch_or))
-      return __scoped_atomic_fetch_or(&val, mask, int(mem_ord),
-                                      (int)(mem_scope));
-    else
-      return __atomic_fetch_or(&val, mask, int(mem_ord));
+#if __has_builtin(__scoped_atomic_fetch_or)
+    return __scoped_atomic_fetch_or(&val, mask, int(mem_ord), (int)(mem_scope));
+#else
+    return __atomic_fetch_or(&val, mask, int(mem_ord));
+#endif
   }
 
   T fetch_and(T mask, MemoryOrder mem_ord = MemoryOrder::SEQ_CST,
               [[maybe_unused]] MemoryScope mem_scope = MemoryScope::DEVICE) {
-    if constexpr (LIBC_HAS_BUILTIN(__scoped_atomic_fetch_and))
-      return __scoped_atomic_fetch_and(&val, mask, int(mem_ord),
-                                       (int)(mem_scope));
-    else
-      return __atomic_fetch_and(&val, mask, int(mem_ord));
+#if __has_builtin(__scoped_atomic_fetch_and)
+    return __scoped_atomic_fetch_and(&val, mask, int(mem_ord),
+                                     (int)(mem_scope));
+#else
+    return __atomic_fetch_and(&val, mask, int(mem_ord));
+#endif
   }
 
   T fetch_sub(T decrement, MemoryOrder mem_ord = MemoryOrder::SEQ_CST,
               [[maybe_unused]] MemoryScope mem_scope = MemoryScope::DEVICE) {
-    if constexpr (LIBC_HAS_BUILTIN(__scoped_atomic_fetch_sub))
-      return __scoped_atomic_fetch_sub(&val, decrement, int(mem_ord),
-                                       (int)(mem_scope));
-    else
-      return __atomic_fetch_sub(&val, decrement, int(mem_ord));
+#if __has_builtin(__scoped_atomic_fetch_sub)
+    return __scoped_atomic_fetch_sub(&val, decrement, int(mem_ord),
+                                     (int)(mem_scope));
+#else
+    return __atomic_fetch_sub(&val, decrement, int(mem_ord));
+#endif
   }
 
   // Set the value without using an atomic operation. This is useful
@@ -166,7 +172,7 @@ LIBC_INLINE void atomic_thread_fence([[maybe_unused]] MemoryOrder mem_ord) {
 // except no instructions for memory ordering are issued. Only reordering of
 // the instructions by the compiler is suppressed as order instructs.
 LIBC_INLINE void atomic_signal_fence([[maybe_unused]] MemoryOrder mem_ord) {
-#if LIBC_HAS_BUILTIN(__atomic_signal_fence)
+#if __has_builtin(__atomic_signal_fence)
   __atomic_signal_fence(static_cast<int>(mem_ord));
 #else
   // if the builtin is not ready, use asm as a full compiler barrier.

diff --git a/libc/src/__support/CPP/bit.h b/libc/src/__support/CPP/bit.h
@@ -14,14 +14,13 @@
 #include "src/__support/CPP/limits.h" // numeric_limits
 #include "src/__support/CPP/type_traits.h"
 #include "src/__support/macros/attributes.h"
-#include "src/__support/macros/config.h" // LIBC_HAS_BUILTIN
 #include "src/__support/macros/sanitizer.h"
 
 #include <stdint.h>
 
 namespace LIBC_NAMESPACE::cpp {
 
-#if LIBC_HAS_BUILTIN(__builtin_memcpy_inline)
+#if __has_builtin(__builtin_memcpy_inline)
 #define LLVM_LIBC_HAS_BUILTIN_MEMCPY_INLINE
 #endif
 
@@ -36,20 +35,20 @@ LIBC_INLINE constexpr cpp::enable_if_t<
     To>
 bit_cast(const From &from) {
   MSAN_UNPOISON(&from, sizeof(From));
-#if LIBC_HAS_BUILTIN(__builtin_bit_cast)
+#if __has_builtin(__builtin_bit_cast)
   return __builtin_bit_cast(To, from);
 #else
   To to;
   char *dst = reinterpret_cast<char *>(&to);
   const char *src = reinterpret_cast<const char *>(&from);
-#if LIBC_HAS_BUILTIN(__builtin_memcpy_inline)
+#if __has_builtin(__builtin_memcpy_inline)
   __builtin_memcpy_inline(dst, src, sizeof(To));
 #else
   for (unsigned i = 0; i < sizeof(To); ++i)
     dst[i] = src[i];
-#endif // LIBC_HAS_BUILTIN(__builtin_memcpy_inline)
+#endif // __has_builtin(__builtin_memcpy_inline)
   return to;
-#endif // LIBC_HAS_BUILTIN(__builtin_bit_cast)
+#endif // __has_builtin(__builtin_bit_cast)
 }
 
 template <typename T>
@@ -94,7 +93,7 @@ countr_zero(T value) {
   }
   return zero_bits;
 }
-#if LIBC_HAS_BUILTIN(__builtin_ctzs)
+#if __has_builtin(__builtin_ctzs)
 ADD_SPECIALIZATION(countr_zero, unsigned short, __builtin_ctzs)
 #endif
 ADD_SPECIALIZATION(countr_zero, unsigned int, __builtin_ctz)
@@ -124,7 +123,7 @@ countl_zero(T value) {
   }
   return zero_bits;
 }
-#if LIBC_HAS_BUILTIN(__builtin_clzs)
+#if __has_builtin(__builtin_clzs)
 ADD_SPECIALIZATION(countl_zero, unsigned short, __builtin_clzs)
 #endif
 ADD_SPECIALIZATION(countl_zero, unsigned int, __builtin_clz)
@@ -242,6 +241,14 @@ LIBC_INLINE constexpr To bit_or_static_cast(const From &from) {
 /// Count number of 1's aka population count or Hamming weight.
 ///
 /// Only unsigned integral types are allowed.
+// clang-19+, gcc-14+
+#if __has_builtin(__builtin_popcountg)
+template <typename T>
+[[nodiscard]] LIBC_INLINE constexpr cpp::enable_if_t<cpp::is_unsigned_v<T>, int>
+popcount(T value) {
+  return __builtin_popcountg(value);
+}
+#else // !__has_builtin(__builtin_popcountg)
 template <typename T>
 [[nodiscard]] LIBC_INLINE constexpr cpp::enable_if_t<cpp::is_unsigned_v<T>, int>
 popcount(T value) {
@@ -261,7 +268,7 @@ ADD_SPECIALIZATION(unsigned short, __builtin_popcount)
 ADD_SPECIALIZATION(unsigned, __builtin_popcount)
 ADD_SPECIALIZATION(unsigned long, __builtin_popcountl)
 ADD_SPECIALIZATION(unsigned long long, __builtin_popcountll)
-// TODO: 128b specializations?
+#endif // __builtin_popcountg
 #undef ADD_SPECIALIZATION
 
 } // namespace LIBC_NAMESPACE::cpp

diff --git a/libc/src/__support/CPP/type_traits/add_pointer.h b/libc/src/__support/CPP/type_traits/add_pointer.h
@@ -10,7 +10,6 @@
 
 #include "src/__support/CPP/type_traits/remove_reference.h"
 #include "src/__support/CPP/type_traits/type_identity.h"
-#include "src/__support/macros/config.h"
 
 namespace LIBC_NAMESPACE::cpp {
 

diff --git a/libc/src/__support/CPP/type_traits/decay.h b/libc/src/__support/CPP/type_traits/decay.h
@@ -9,7 +9,6 @@
 #define LLVM_LIBC_SRC___SUPPORT_CPP_TYPE_TRAITS_DECAY_H
 
 #include "src/__support/macros/attributes.h"
-#include "src/__support/macros/config.h"
 
 #include "src/__support/CPP/type_traits/add_pointer.h"
 #include "src/__support/CPP/type_traits/conditional.h"

diff --git a/libc/src/__support/CPP/type_traits/is_destructible.h b/libc/src/__support/CPP/type_traits/is_destructible.h
@@ -16,12 +16,11 @@
 #include "src/__support/CPP/type_traits/true_type.h"
 #include "src/__support/CPP/type_traits/type_identity.h"
 #include "src/__support/macros/attributes.h"
-#include "src/__support/macros/config.h"
 
 namespace LIBC_NAMESPACE::cpp {
 
 // is_destructible
-#if LIBC_HAS_BUILTIN(__is_destructible)
+#if __has_builtin(__is_destructible)
 template <typename T>
 struct is_destructible : bool_constant<__is_destructible(T)> {};
 #else

diff --git a/libc/src/__support/CPP/type_traits/is_function.h b/libc/src/__support/CPP/type_traits/is_function.h
@@ -12,12 +12,11 @@
 #include "src/__support/CPP/type_traits/is_const.h"
 #include "src/__support/CPP/type_traits/is_reference.h"
 #include "src/__support/macros/attributes.h"
-#include "src/__support/macros/config.h"
 
 namespace LIBC_NAMESPACE::cpp {
 
 // is_function
-#if LIBC_HAS_BUILTIN(__is_function)
+#if __has_builtin(__is_function)
 template <typename T>
 struct is_function : integral_constant<bool, __is_function(T)> {};
 #else

diff --git a/libc/src/__support/CPP/type_traits/is_lvalue_reference.h b/libc/src/__support/CPP/type_traits/is_lvalue_reference.h
@@ -12,12 +12,11 @@
 #include "src/__support/CPP/type_traits/false_type.h"
 #include "src/__support/CPP/type_traits/true_type.h"
 #include "src/__support/macros/attributes.h"
-#include "src/__support/macros/config.h"
 
 namespace LIBC_NAMESPACE::cpp {
 
 // is_lvalue_reference
-#if LIBC_HAS_BUILTIN(__is_lvalue_reference)
+#if __has_builtin(__is_lvalue_reference)
 template <typename T>
 struct is_lvalue_reference : bool_constant<__is_lvalue_reference(T)> {};
 #else

diff --git a/libc/src/__support/CPP/type_traits/is_reference.h b/libc/src/__support/CPP/type_traits/is_reference.h
@@ -12,12 +12,11 @@
 #include "src/__support/CPP/type_traits/false_type.h"
 #include "src/__support/CPP/type_traits/true_type.h"
 #include "src/__support/macros/attributes.h"
-#include "src/__support/macros/config.h"
 
 namespace LIBC_NAMESPACE::cpp {
 
 // is_reference
-#if LIBC_HAS_BUILTIN(__is_reference)
+#if __has_builtin(__is_reference)
 template <typename T> struct is_reference : bool_constant<__is_reference(T)> {};
 #else
 template <typename T> struct is_reference : public false_type {};

diff --git a/libc/src/__support/CPP/type_traits/is_rvalue_reference.h b/libc/src/__support/CPP/type_traits/is_rvalue_reference.h
@@ -12,12 +12,11 @@
 #include "src/__support/CPP/type_traits/false_type.h"
 #include "src/__support/CPP/type_traits/true_type.h"
 #include "src/__support/macros/attributes.h"
-#include "src/__support/macros/config.h"
 
 namespace LIBC_NAMESPACE::cpp {
 
 // is_rvalue_reference
-#if LIBC_HAS_BUILTIN(__is_rvalue_reference)
+#if __has_builtin(__is_rvalue_reference)
 template <typename T>
 struct is_rvalue_reference : bool_constant<__is_rvalue_reference(T)> {};
 #else

diff --git a/libc/src/__support/CPP/type_traits/is_trivially_copyable.h b/libc/src/__support/CPP/type_traits/is_trivially_copyable.h
@@ -9,7 +9,6 @@
 #define LLVM_LIBC_SRC___SUPPORT_CPP_TYPE_TRAITS_IS_TRIVIALLY_COPYABLE_H
 
 #include "src/__support/CPP/type_traits/integral_constant.h"
-#include "src/__support/macros/config.h"
 
 namespace LIBC_NAMESPACE::cpp {
 

diff --git a/libc/src/__support/CPP/type_traits/is_trivially_destructible.h b/libc/src/__support/CPP/type_traits/is_trivially_destructible.h
@@ -11,12 +11,11 @@
 #include "src/__support/CPP/type_traits/bool_constant.h"
 #include "src/__support/CPP/type_traits/is_destructible.h"
 #include "src/__support/macros/attributes.h"
-#include "src/__support/macros/config.h"
 
 namespace LIBC_NAMESPACE::cpp {
 
 // is_trivially_destructible
-#if LIBC_HAS_BUILTIN(__is_trivially_destructible)
+#if __has_builtin(__is_trivially_destructible)
 template <typename T>
 struct is_trivially_destructible
     : public bool_constant<__is_trivially_destructible(T)> {};
@@ -25,7 +24,7 @@ template <typename T>
 struct is_trivially_destructible
     : public bool_constant<cpp::is_destructible_v<T> &&__has_trivial_destructor(
           T)> {};
-#endif // LIBC_HAS_BUILTIN(__is_trivially_destructible)
+#endif // __has_builtin(__is_trivially_destructible)
 template <typename T>
 LIBC_INLINE_VAR constexpr bool is_trivially_destructible_v =
     is_trivially_destructible<T>::value;

diff --git a/libc/src/__support/CPP/type_traits/remove_all_extents.h b/libc/src/__support/CPP/type_traits/remove_all_extents.h
@@ -9,14 +9,13 @@
 #define LLVM_LIBC_SRC___SUPPORT_CPP_TYPE_TRAITS_REMOVE_ALL_EXTENTS_H
 
 #include "src/__support/CPP/type_traits/type_identity.h"
-#include "src/__support/macros/config.h"
 
 #include <stddef.h> // size_t
 
 namespace LIBC_NAMESPACE::cpp {
 
 // remove_all_extents
-#if LIBC_HAS_BUILTIN(__remove_all_extents)
+#if __has_builtin(__remove_all_extents)
 template <typename T> using remove_all_extents_t = __remove_all_extents(T);
 template <typename T>
 struct remove_all_extents : cpp::type_identity<remove_all_extents_t<T>> {};

diff --git a/libc/src/__support/FPUtil/CMakeLists.txt b/libc/src/__support/FPUtil/CMakeLists.txt
@@ -6,7 +6,6 @@ add_header_library(
     libc.include.fenv
     libc.include.math
     libc.src.__support.macros.attributes
-    libc.src.__support.macros.config
     libc.src.errno.errno
 )
 

diff --git a/libc/src/__support/FPUtil/FEnvImpl.h b/libc/src/__support/FPUtil/FEnvImpl.h
@@ -11,7 +11,6 @@
 
 #include "include/llvm-libc-macros/math-macros.h"
 #include "src/__support/macros/attributes.h" // LIBC_INLINE
-#include "src/__support/macros/config.h"     // LIBC_HAS_BUILTIN
 #include "src/__support/macros/properties/architectures.h"
 #include "src/errno/libc_errno.h"
 #include <fenv.h>

diff --git a/libc/src/__support/FPUtil/gpu/FMA.h b/libc/src/__support/FPUtil/gpu/FMA.h
@@ -10,12 +10,12 @@
 #define LLVM_LIBC_SRC___SUPPORT_FPUTIL_GPU_FMA_H
 
 #include "src/__support/CPP/type_traits.h"
-#include "src/__support/macros/config.h"
 
-// These intrinsics map to the FMA instrunctions in the target ISA for the GPU.
+// These intrinsics map to the FMA instructions in the target ISA for the GPU.
 // The default rounding mode generated from these will be to the nearest even.
-static_assert(LIBC_HAS_BUILTIN(__builtin_fma), "FMA builtins must be defined");
-static_assert(LIBC_HAS_BUILTIN(__builtin_fmaf), "FMA builtins must be defined");
+#if !__has_builtin(__builtin_fma) || !__has_builtin(__builtin_fmaf)
+#error "FMA builtins must be defined");
+#endif
 
 namespace LIBC_NAMESPACE {
 namespace fputil {

diff --git a/libc/src/__support/UInt.h b/libc/src/__support/UInt.h
@@ -1082,6 +1082,17 @@ bit_cast(const UInt<Bits> &from) {
   return cpp::bit_cast<To>(from.val);
 }
 
+// Specialization of cpp::popcount ('bit.h') for BigInt.
+template <typename T>
+[[nodiscard]] LIBC_INLINE constexpr cpp::enable_if_t<is_big_int_v<T>, int>
+popcount(T value) {
+  int bits = 0;
+  for (auto word : value.val)
+    if (word)
+      bits += popcount(word);
+  return bits;
+}
+
 // Specialization of cpp::has_single_bit ('bit.h') for BigInt.
 template <typename T>
 [[nodiscard]] LIBC_INLINE constexpr cpp::enable_if_t<is_big_int_v<T>, bool>
@@ -1218,6 +1229,49 @@ LIBC_INLINE constexpr cpp::enable_if_t<is_big_int_v<T>, T> mask_leading_ones() {
   return out;
 }
 
+// Specialization of count_zeros ('math_extras.h') for BigInt.
+template <typename T>
+[[nodiscard]]
+LIBC_INLINE constexpr cpp::enable_if_t<is_big_int_v<T>, int>
+count_zeros(T value) {
+  return cpp::popcount(~value);
+}
+
+// Specialization of first_leading_zero ('math_extras.h') for BigInt.
+template <typename T>
+[[nodiscard]]
+LIBC_INLINE constexpr cpp::enable_if_t<is_big_int_v<T>, int>
+first_leading_zero(T value) {
+  return value == cpp::numeric_limits<T>::max() ? 0
+                                                : cpp::countl_one(value) + 1;
+}
+
+// Specialization of first_leading_one ('math_extras.h') for BigInt.
+template <typename T>
+[[nodiscard]]
+LIBC_INLINE constexpr cpp::enable_if_t<is_big_int_v<T>, int>
+first_leading_one(T value) {
+  return first_leading_zero(~value);
+}
+
+// Specialization of first_trailing_zero ('math_extras.h') for BigInt.
+template <typename T>
+[[nodiscard]]
+LIBC_INLINE constexpr cpp::enable_if_t<is_big_int_v<T>, int>
+first_trailing_zero(T value) {
+  return value == cpp::numeric_limits<T>::max() ? 0
+                                                : cpp::countr_zero(~value) + 1;
+}
+
+// Specialization of first_trailing_one ('math_extras.h') for BigInt.
+template <typename T>
+[[nodiscard]]
+LIBC_INLINE constexpr cpp::enable_if_t<is_big_int_v<T>, int>
+first_trailing_one(T value) {
+  return value == cpp::numeric_limits<T>::max() ? 0
+                                                : cpp::countr_zero(value) + 1;
+}
+
 } // namespace LIBC_NAMESPACE
 
 #endif // LLVM_LIBC_SRC___SUPPORT_UINT_H
diff --git a/libc/src/__support/macros/config.h b/libc/src/__support/macros/config.h
@@ -13,24 +13,6 @@
 #ifndef LLVM_LIBC_SRC___SUPPORT_MACROS_CONFIG_H
 #define LLVM_LIBC_SRC___SUPPORT_MACROS_CONFIG_H
 
-// LIBC_HAS_BUILTIN()
-//
-// Checks whether the compiler supports a Clang Feature Checking Macro, and if
-// so, checks whether it supports the provided builtin function "x" where x
-// is one of the functions noted in
-// https://clang.llvm.org/docs/LanguageExtensions.html
-//
-// Note: Use this macro to avoid an extra level of #ifdef __has_builtin check.
-// http://releases.llvm.org/3.3/tools/clang/docs/LanguageExtensions.html
-
-// Compiler builtin-detection.
-// clang.llvm.org/docs/LanguageExtensions.html#has-builtin
-#ifdef __has_builtin
-#define LIBC_HAS_BUILTIN(x) __has_builtin(x)
-#else
-#define LIBC_HAS_BUILTIN(x) 0
-#endif
-
 // Compiler feature-detection.
 // clang.llvm.org/docs/LanguageExtensions.html#has-feature-and-has-extension
 #ifdef __has_feature

diff --git a/libc/src/__support/macros/optimization.h b/libc/src/__support/macros/optimization.h
@@ -11,7 +11,6 @@
 #define LLVM_LIBC_SRC___SUPPORT_MACROS_OPTIMIZATION_H
 
 #include "src/__support/macros/attributes.h"          // LIBC_INLINE
-#include "src/__support/macros/config.h"              // LIBC_HAS_BUILTIN
 #include "src/__support/macros/properties/compiler.h" // LIBC_COMPILER_IS_CLANG
 
 // We use a template to implement likely/unlikely to make sure that we don't

diff --git a/libc/src/__support/macros/sanitizer.h b/libc/src/__support/macros/sanitizer.h
@@ -47,8 +47,7 @@
 // Functions to unpoison memory
 //-----------------------------------------------------------------------------
 
-#if defined(LIBC_HAVE_MEMORY_SANITIZER) &&                                     \
-    LIBC_HAS_BUILTIN(__builtin_constant_p)
+#if defined(LIBC_HAVE_MEMORY_SANITIZER) && __has_builtin(__builtin_constant_p)
 // Only perform MSAN unpoison in non-constexpr context.
 #include <sanitizer/msan_interface.h>
 #define MSAN_UNPOISON(addr, size)                                              \

diff --git a/libc/src/__support/math_extras.h b/libc/src/__support/math_extras.h
@@ -14,7 +14,6 @@
 #include "src/__support/CPP/limits.h"        // CHAR_BIT, numeric_limits
 #include "src/__support/CPP/type_traits.h"   // is_unsigned_v
 #include "src/__support/macros/attributes.h" // LIBC_INLINE
-#include "src/__support/macros/config.h"     // LIBC_HAS_BUILTIN
 
 namespace LIBC_NAMESPACE {
 
@@ -61,7 +60,7 @@ add_with_carry(T a, T b, T carry_in) {
   return add_with_carry_const<T>(a, b, carry_in);
 }
 
-#if LIBC_HAS_BUILTIN(__builtin_addc)
+#if __has_builtin(__builtin_addc)
 // https://clang.llvm.org/docs/LanguageExtensions.html#multiprecision-arithmetic-builtins
 
 template <>
@@ -129,7 +128,7 @@ add_with_carry<unsigned long long>(unsigned long long a, unsigned long long b,
   }
 }
 
-#endif // LIBC_HAS_BUILTIN(__builtin_addc)
+#endif // __has_builtin(__builtin_addc)
 
 // Subtract with borrow
 template <typename T> struct DiffBorrow {
@@ -157,7 +156,7 @@ sub_with_borrow(T a, T b, T borrow_in) {
   return sub_with_borrow_const<T>(a, b, borrow_in);
 }
 
-#if LIBC_HAS_BUILTIN(__builtin_subc)
+#if __has_builtin(__builtin_subc)
 // https://clang.llvm.org/docs/LanguageExtensions.html#multiprecision-arithmetic-builtins
 
 template <>
@@ -225,7 +224,7 @@ sub_with_borrow<unsigned long long>(unsigned long long a, unsigned long long b,
   }
 }
 
-#endif // LIBC_HAS_BUILTIN(__builtin_subc)
+#endif // __has_builtin(__builtin_subc)
 
 template <typename T>
 [[nodiscard]] LIBC_INLINE constexpr cpp::enable_if_t<cpp::is_unsigned_v<T>, int>

diff --git a/libc/src/__support/memory_size.h b/libc/src/__support/memory_size.h
@@ -19,7 +19,7 @@
 namespace LIBC_NAMESPACE {
 namespace internal {
 template <class T> LIBC_INLINE bool mul_overflow(T a, T b, T *res) {
-#if LIBC_HAS_BUILTIN(__builtin_mul_overflow)
+#if __has_builtin(__builtin_mul_overflow)
   return __builtin_mul_overflow(a, b, res);
 #else
   T max = cpp::numeric_limits<T>::max();

diff --git a/libc/src/string/memory_utils/generic/builtin.h b/libc/src/string/memory_utils/generic/builtin.h
@@ -10,16 +10,16 @@
 #define LLVM_LIBC_SRC_STRING_MEMORY_UTILS_GENERIC_BUILTIN_H
 
 #include "src/__support/macros/attributes.h" // LIBC_INLINE
-#include "src/__support/macros/config.h"     // LIBC_HAS_BUILTIN
 #include "src/string/memory_utils/utils.h"   // Ptr, CPtr
 
 #include <stddef.h> // size_t
 
 namespace LIBC_NAMESPACE {
 
-static_assert(LIBC_HAS_BUILTIN(__builtin_memcpy), "Builtin not defined");
-static_assert(LIBC_HAS_BUILTIN(__builtin_memset), "Builtin not defined");
-static_assert(LIBC_HAS_BUILTIN(__builtin_memmove), "Builtin not defined");
+#if !__has_builtin(__builtin_memcpy) || !__has_builtin(__builtin_memset) ||    \
+    !__has_builtin(__builtin_memmove)
+#error "Builtin not defined");
+#endif
 
 [[maybe_unused]] LIBC_INLINE void
 inline_memcpy_builtin(Ptr dst, CPtr src, size_t count, size_t offset = 0) {