[clang] [llvm] [WIP] Correct lowering of `fp128` intrinsics (PR #76558)

Trevor Gross via cfe-commits cfe-commits at lists.llvm.org
Fri Dec 29 16:32:15 PST 2023


https://github.com/tgross35 updated https://github.com/llvm/llvm-project/pull/76558

>From 590f4920ceb1a80d711d39624b0249cd9ff774d2 Mon Sep 17 00:00:00 2001
From: Trevor Gross <tmgross at umich.edu>
Date: Fri, 11 Aug 2023 22:16:01 -0400
Subject: [PATCH 1/4] [IR] Add an xpassing test for `f128` intrinsic lowering

`f128` intrinsic functions lower to incorrect libc calls. Add a test
showing current behavior.
---
 .../CodeGen/Generic/f128-math-lowering.ll     | 610 ++++++++++++++++++
 1 file changed, 610 insertions(+)
 create mode 100644 llvm/test/CodeGen/Generic/f128-math-lowering.ll

diff --git a/llvm/test/CodeGen/Generic/f128-math-lowering.ll b/llvm/test/CodeGen/Generic/f128-math-lowering.ll
new file mode 100644
index 00000000000000..8a70786d97fe67
--- /dev/null
+++ b/llvm/test/CodeGen/Generic/f128-math-lowering.ll
@@ -0,0 +1,610 @@
+
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 2
+;
+; RUN: llc < %s -mtriple=aarch64-unknown-unknown -verify-machineinstrs | FileCheck %s --check-prefix=CHECK-AARCH64
+; RUN: llc < %s -mtriple=riscv32-unknown-unknown -verify-machineinstrs | FileCheck %s --check-prefix=CHECK-RISCV32
+; RUN: llc < %s -mtriple=s390x-unknown-unknown -verify-machineinstrs | FileCheck %s --check-prefix=CHECK-S390X
+; RUN: llc < %s -mtriple=i686-unknown-unknown   -verify-machineinstrs | FileCheck %s --check-prefix=CHECK-X86
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -verify-machineinstrs | FileCheck %s --check-prefix=CHECK-X64
+;
+; Verify that fp128 intrinsics only lower to `long double` calls on platforms
+; where `f128` and `long double` have the same layout.
+;
+; We test on x86 and x64 which have 80-bit ld, as well as aarch64 (ld == f128),
+; riscv32 (ld == f64), and s380x (ld == f128 with different alignment from
+; x64/aarch64 f128).
+;
+; FIXME: these emit calls to long double functions but should emit f128 calls
+
+define fp128 @test_cbrtf128(fp128 %a) {
+; CHECK-LABEL:      test_cbrtf128:
+; CHECK-AARCH64:    b llvm.cbrt.f128
+; CHECK-RISCV32:    call llvm.cbrt.f128 at plt
+; CHECK-S390X:      brasl {{%.*}} llvm.cbrt.f128 at PLT
+; CHECK-X64:        jmp llvm.cbrt.f128 at PLT # TAILCALL
+; CHECK-X86:        calll llvm.cbrt.f128 at PLT
+start:
+  %0 = tail call fp128 @llvm.cbrt.f128(fp128 %a)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.cbrt.f128(fp128)
+
+
+define fp128 @test_ceilf128(fp128 %a) {
+; CHECK-LABEL:      test_ceilf128:
+; CHECK-AARCH64:    b ceill
+; CHECK-RISCV32:    call ceill at plt
+; CHECK-S390X:      brasl {{%.*}} ceill at PLT
+; CHECK-X64:        jmp ceill at PLT
+; CHECK-X86:        calll ceill
+start:
+  %0 = tail call fp128 @llvm.ceil.f128(fp128 %a)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.ceil.f128(fp128)
+
+
+define fp128 @test_copysignf128(fp128 %a, fp128 %b) {
+; No math library call here, so make sure the assembly does the correct thing.
+; This test is autogenerated
+; CHECK-LABEL:        test_copysignf128:
+; CHECK-AARCH64-LABEL: test_copysignf128:
+; CHECK-AARCH64:       // %bb.0: // %start
+; CHECK-AARCH64-NEXT:    stp q0, q1, [sp, #-32]!
+; CHECK-AARCH64-NEXT:    .cfi_def_cfa_offset 32
+; CHECK-AARCH64-NEXT:    ldrb w8, [sp, #15]
+; CHECK-AARCH64-NEXT:    ldrb w9, [sp, #31]
+; CHECK-AARCH64-NEXT:    bfxil w9, w8, #0, #7
+; CHECK-AARCH64-NEXT:    strb w9, [sp, #15]
+; CHECK-AARCH64-NEXT:    ldr q0, [sp], #32
+; CHECK-AARCH64-NEXT:    ret
+;
+; CHECK-RISCV32-LABEL: test_copysignf128:
+; CHECK-RISCV32:       # %bb.0: # %start
+; CHECK-RISCV32-NEXT:    lw a3, 0(a1)
+; CHECK-RISCV32-NEXT:    lw a4, 4(a1)
+; CHECK-RISCV32-NEXT:    lw a2, 12(a2)
+; CHECK-RISCV32-NEXT:    lw a5, 12(a1)
+; CHECK-RISCV32-NEXT:    lw a1, 8(a1)
+; CHECK-RISCV32-NEXT:    lui a6, 524288
+; CHECK-RISCV32-NEXT:    and a2, a2, a6
+; CHECK-RISCV32-NEXT:    slli a5, a5, 1
+; CHECK-RISCV32-NEXT:    srli a5, a5, 1
+; CHECK-RISCV32-NEXT:    or a2, a5, a2
+; CHECK-RISCV32-NEXT:    sw a1, 8(a0)
+; CHECK-RISCV32-NEXT:    sw a4, 4(a0)
+; CHECK-RISCV32-NEXT:    sw a3, 0(a0)
+; CHECK-RISCV32-NEXT:    sw a2, 12(a0)
+; CHECK-RISCV32-NEXT:    ret
+;
+; CHECK-S390X-LABEL: test_copysignf128:
+; CHECK-S390X:       # %bb.0: # %start
+; CHECK-S390X-NEXT:    ld %f0, 0(%r3)
+; CHECK-S390X-NEXT:    ld %f2, 8(%r3)
+; CHECK-S390X-NEXT:    ld %f1, 0(%r4)
+; CHECK-S390X-NEXT:    ld %f3, 8(%r4)
+; CHECK-S390X-NEXT:    cpsdr %f0, %f1, %f0
+; CHECK-S390X-NEXT:    std %f0, 0(%r2)
+; CHECK-S390X-NEXT:    std %f2, 8(%r2)
+; CHECK-S390X-NEXT:    br %r14
+;
+; CHECK-X86-LABEL: test_copysignf128:
+; CHECK-X86:       # %bb.0: # %start
+; CHECK-X86-NEXT:    pushl %ebx
+; CHECK-X86-NEXT:    .cfi_def_cfa_offset 8
+; CHECK-X86-NEXT:    pushl %edi
+; CHECK-X86-NEXT:    .cfi_def_cfa_offset 12
+; CHECK-X86-NEXT:    pushl %esi
+; CHECK-X86-NEXT:    .cfi_def_cfa_offset 16
+; CHECK-X86-NEXT:    .cfi_offset %esi, -16
+; CHECK-X86-NEXT:    .cfi_offset %edi, -12
+; CHECK-X86-NEXT:    .cfi_offset %ebx, -8
+; CHECK-X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; CHECK-X86-NEXT:    movl {{[0-9]+}}(%esp), %ecx
+; CHECK-X86-NEXT:    movl {{[0-9]+}}(%esp), %edx
+; CHECK-X86-NEXT:    movl {{[0-9]+}}(%esp), %esi
+; CHECK-X86-NEXT:    movl $-2147483648, %edi # imm = 0x80000000
+; CHECK-X86-NEXT:    andl {{[0-9]+}}(%esp), %edi
+; CHECK-X86-NEXT:    movl $2147483647, %ebx # imm = 0x7FFFFFFF
+; CHECK-X86-NEXT:    andl {{[0-9]+}}(%esp), %ebx
+; CHECK-X86-NEXT:    orl %edi, %ebx
+; CHECK-X86-NEXT:    movl %ebx, 12(%eax)
+; CHECK-X86-NEXT:    movl %esi, 8(%eax)
+; CHECK-X86-NEXT:    movl %edx, 4(%eax)
+; CHECK-X86-NEXT:    movl %ecx, (%eax)
+; CHECK-X86-NEXT:    popl %esi
+; CHECK-X86-NEXT:    .cfi_def_cfa_offset 12
+; CHECK-X86-NEXT:    popl %edi
+; CHECK-X86-NEXT:    .cfi_def_cfa_offset 8
+; CHECK-X86-NEXT:    popl %ebx
+; CHECK-X86-NEXT:    .cfi_def_cfa_offset 4
+; CHECK-X86-NEXT:    retl $4
+;
+; CHECK-X64-LABEL: test_copysignf128:
+; CHECK-X64:       # %bb.0: # %start
+; CHECK-X64-NEXT:    andps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1
+; CHECK-X64-NEXT:    andps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
+; CHECK-X64-NEXT:    orps %xmm1, %xmm0
+; CHECK-X64-NEXT:    retq
+start:
+  %0 = tail call fp128 @llvm.copysign.f128(fp128 %a, fp128 %b)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.copysign.f128(fp128, fp128)
+
+
+define fp128 @test_cosf128(fp128 %a) {
+; CHECK-LABEL:      test_cosf128:
+; CHECK-AARCH64:    b cosl
+; CHECK-RISCV32:    call cosl at plt
+; CHECK-S390X:      brasl {{%.*}} cosl at PLT
+; CHECK-X64:        jmp cosl at PLT
+; CHECK-X86:        calll cosl
+start:
+  %0 = tail call fp128 @llvm.cos.f128(fp128 %a)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.cos.f128(fp128)
+
+
+define fp128 @test_exp2f128(fp128 %a) {
+; CHECK-LABEL:      test_exp2f128:
+; CHECK-AARCH64:    b exp2l
+; CHECK-RISCV32:    call exp2l at plt
+; CHECK-S390X:      brasl {{%.*}} exp2l at PLT
+; CHECK-X64:        jmp exp2l at PLT
+; CHECK-X86:        calll exp2l
+start:
+  %0 = tail call fp128 @llvm.exp2.f128(fp128 %a)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.exp2.f128(fp128)
+
+
+define fp128 @test___exp2f128_finite(fp128 %a) {
+; CHECK-LABEL:      test___exp2f128_finite:
+; CHECK-AARCH64:    b llvm.__exp2f128_finite.f128
+; CHECK-RISCV32:    call llvm.__exp2f128_finite.f128 at plt
+; CHECK-S390X:      brasl {{%.*}} llvm.__exp2f128_finite.f128 at PLT
+; CHECK-X64:        jmp llvm.__exp2f128_finite.f128 at PLT # TAILCALL
+; CHECK-X86:        calll llvm.__exp2f128_finite.f128 at PLT
+start:
+  %0 = tail call fp128 @llvm.__exp2f128_finite.f128(fp128  %a)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.__exp2f128_finite.f128(fp128)
+
+
+define fp128 @test_expf128(fp128 %a) {
+; CHECK-LABEL:      test_expf128:
+; CHECK-AARCH64:    b expl
+; CHECK-RISCV32:    call expl at plt
+; CHECK-S390X:      brasl {{%.*}} expl at PLT
+; CHECK-X64:        jmp expl at PLT
+; CHECK-X86:        calll expl
+start:
+  %0 = tail call fp128 @llvm.exp.f128(fp128 %a)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.exp.f128(fp128)
+
+
+define fp128 @test___expf128_finite(fp128 %a) {
+; CHECK-LABEL:      test___expf128_finite:
+; CHECK-AARCH64:    b llvm.__expf128_finite.f128
+; CHECK-RISCV32:    call llvm.__expf128_finite.f128 at plt
+; CHECK-S390X:      brasl {{%.*}} llvm.__expf128_finite.f128 at PLT
+; CHECK-X64:        jmp llvm.__expf128_finite.f128 at PLT # TAILCALL
+; CHECK-X86:        calll llvm.__expf128_finite.f128 at PLT
+start:
+  %0 = tail call fp128 @llvm.__expf128_finite.f128(fp128 %a)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.__expf128_finite.f128(fp128)
+
+
+define fp128 @test_floorf128(fp128 %a) {
+; CHECK-LABEL:      test_floorf128:
+; CHECK-AARCH64:    b floorl
+; CHECK-RISCV32:    call floorl at plt
+; CHECK-S390X:      brasl {{%.*}} floorl at PLT
+; CHECK-X64:        jmp floorl at PLT
+; CHECK-X86:        calll floorl
+start:
+  %0 = tail call fp128 @llvm.floor.f128(fp128 %a)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.floor.f128(fp128)
+
+
+define fp128 @test_fmaf128(fp128 %a, fp128 %b, fp128 %c) {
+; CHECK-LABEL:      test_fmaf128:
+; CHECK-AARCH64:    b fmal
+; CHECK-RISCV32:    call fmal at plt
+; CHECK-S390X:      brasl {{%.*}} fmal at PLT
+; CHECK-X64:        jmp fmal at PLT
+; CHECK-X86:        calll fmal
+start:
+  %0 = tail call fp128 @llvm.fma.f128(fp128 %a, fp128 %b, fp128 %c)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.fma.f128(fp128, fp128, fp128)
+
+
+define fp128 @test_fmaxf128(fp128 %a, fp128 %b) {
+; CHECK-LABEL:      test_fmaxf128:
+; CHECK-AARCH64:    b llvm.fmax.f128
+; CHECK-RISCV32:    call llvm.fmax.f128 at plt
+; CHECK-S390X:      brasl {{%.*}} llvm.fmax.f128 at PLT
+; CHECK-X64:        jmp llvm.fmax.f128 at PLT # TAILCALL
+; CHECK-X86:        calll llvm.fmax.f128 at PLT
+start:
+  %0 = tail call fp128 @llvm.fmax.f128(fp128 %a, fp128 %b)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.fmax.f128(fp128, fp128)
+
+
+define fp128 @test_fminf128(fp128 %a, fp128 %b) {
+; CHECK-LABEL:      test_fminf128:
+; CHECK-AARCH64:    b llvm.fmin.f128
+; CHECK-RISCV32:    call llvm.fmin.f128 at plt
+; CHECK-S390X:      brasl {{%.*}} llvm.fmin.f128 at PLT
+; CHECK-X64:        jmp llvm.fmin.f128 at PLT # TAILCALL
+; CHECK-X86:        calll llvm.fmin.f128 at PLT
+start:
+  %0 = tail call fp128 @llvm.fmin.f128(fp128 %a, fp128 %b)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.fmin.f128(fp128, fp128)
+
+
+define fp128 @test_fmodf128(fp128 %a, fp128 %b) {
+; CHECK-LABEL:      test_fmodf128:
+; CHECK-AARCH64:    b llvm.fmod.f128
+; CHECK-RISCV32:    call llvm.fmod.f128 at plt
+; CHECK-S390X:      brasl {{%.*}} llvm.fmod.f128 at PLT
+; CHECK-X64:        jmp llvm.fmod.f128 at PLT # TAILCALL
+; CHECK-X86:        calll llvm.fmod.f128 at PLT
+start:
+  %0 = tail call fp128 @llvm.fmod.f128(fp128 %a, fp128 %b)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.fmod.f128(fp128, fp128)
+
+
+define { fp128, i32 } @test_frexpf128(fp128 %a) {
+; CHECK-LABEL:      test_frexpf128:
+; CHECK-AARCH64:    bl frexpl
+; CHECK-RISCV32:    call frexpl at plt
+; CHECK-S390X:      brasl {{%.*}} frexpl at PLT
+; CHECK-X64:        callq frexpl at PLT
+; CHECK-X86:        calll frexpl
+start:
+  %0 = tail call { fp128, i32 } @llvm.frexp.f128(fp128 %a)
+  ret { fp128, i32 } %0
+}
+
+declare { fp128, i32 } @llvm.frexp.f128(fp128)
+
+
+define fp128 @test_ldexpf128(fp128 %a, i32 %b) {
+; CHECK-LABEL:      test_ldexpf128:
+; CHECK-AARCH64:    b ldexpl
+; CHECK-RISCV32:    call ldexpl at plt
+; CHECK-S390X:      brasl {{%.*}} ldexpl at PLT
+; CHECK-X64:        jmp ldexpl at PLT
+; CHECK-X86:        calll ldexpl
+start:
+  %0 = tail call fp128 @llvm.ldexp.f128(fp128 %a, i32 %b)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.ldexp.f128(fp128, i32)
+
+
+define i64 @test_llrintf128(fp128 %a) {
+; CHECK-LABEL:      test_llrintf128:
+; CHECK-AARCH64:    b llrintl
+; CHECK-RISCV32:    call llrintl at plt
+; CHECK-S390X:      brasl {{%.*}} llrintl at PLT
+; CHECK-X64:        jmp llrintl at PLT
+; CHECK-X86:        calll llrintl
+start:
+  %0 = tail call i64 @llvm.llrint.f128(fp128 %a)
+  ret i64 %0
+}
+
+declare i64 @llvm.llrint.f128(fp128)
+
+
+define i64 @test_llroundf128(fp128 %a) {
+; CHECK-LABEL:      test_llroundf128:
+; CHECK-AARCH64:    b llroundl
+; CHECK-RISCV32:    call llroundl at plt
+; CHECK-S390X:      brasl {{%.*}} llroundl at PLT
+; CHECK-X64:        jmp llroundl at PLT
+; CHECK-X86:        calll llroundl
+start:
+  %0 = tail call i64 @llvm.llround.i64.f128(fp128 %a)
+  ret i64 %0
+}
+
+declare i64 @llvm.llround.i64.f128(fp128)
+
+
+define fp128 @test_log10f128(fp128 %a) {
+; CHECK-LABEL:      test_log10f128:
+; CHECK-AARCH64:    b log10l
+; CHECK-RISCV32:    call log10l at plt
+; CHECK-S390X:      brasl {{%.*}} log10l at PLT
+; CHECK-X64:        jmp log10l at PLT
+; CHECK-X86:        calll log10l
+start:
+  %0 = tail call fp128 @llvm.log10.f128(fp128 %a)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.log10.f128(fp128)
+
+
+define fp128 @test___log10f128_finite(fp128 %a) {
+; CHECK-LABEL:      test___log10f128_finite:
+; CHECK-AARCH64:    b llvm.__log10f128_finite.f128
+; CHECK-RISCV32:    call llvm.__log10f128_finite.f128 at plt
+; CHECK-S390X:      brasl {{%.*}} llvm.__log10f128_finite.f128 at PLT
+; CHECK-X64:        jmp llvm.__log10f128_finite.f128 at PLT # TAILCALL
+; CHECK-X86:        calll llvm.__log10f128_finite.f128 at PLT
+start:
+  %0 = tail call fp128 @llvm.__log10f128_finite.f128(fp128 %a)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.__log10f128_finite.f128(fp128)
+
+
+define fp128 @test_log2f128(fp128 %a) {
+; CHECK-LABEL:      test_log2f128:
+; CHECK-AARCH64:    b log2l
+; CHECK-RISCV32:    call log2l at plt
+; CHECK-S390X:      brasl {{%.*}} log2l at PLT
+; CHECK-X64:        jmp log2l at PLT
+; CHECK-X86:        calll log2l
+start:
+  %0 = tail call fp128 @llvm.log2.f128(fp128 %a)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.log2.f128(fp128)
+
+
+define fp128 @test___log2f128_finite(fp128 %a) {
+; CHECK-LABEL:      test___log2f128_finite:
+; CHECK-AARCH64:    b llvm.__log2f128_finite.f128
+; CHECK-RISCV32:    call llvm.__log2f128_finite.f128 at plt
+; CHECK-S390X:      brasl {{%.*}} llvm.__log2f128_finite.f128 at PLT
+; CHECK-X64:        jmp llvm.__log2f128_finite.f128 at PLT # TAILCALL
+; CHECK-X86:        calll llvm.__log2f128_finite.f128 at PLT
+start:
+  %0 = tail call fp128 @llvm.__log2f128_finite.f128(fp128 %a)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.__log2f128_finite.f128(fp128)
+
+
+define fp128 @test_logf128(fp128 %a) {
+; CHECK-LABEL:      test_logf128:
+; CHECK-AARCH64:    b logl
+; CHECK-RISCV32:    call logl at plt
+; CHECK-S390X:      brasl {{%.*}} logl at PLT
+; CHECK-X64:        jmp logl at PLT
+; CHECK-X86:        calll logl
+start:
+  %0 = tail call fp128 @llvm.log.f128(fp128 %a)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.log.f128(fp128)
+
+
+define fp128 @test___logf128_finite(fp128 %a) {
+; CHECK-LABEL:      test___logf128_finite:
+; CHECK-AARCH64:    b llvm.__logf128_finite.f128
+; CHECK-RISCV32:    call llvm.__logf128_finite.f128 at plt
+; CHECK-S390X:      brasl {{%.*}} llvm.__logf128_finite.f128 at PLT
+; CHECK-X64:        jmp llvm.__logf128_finite.f128 at PLT # TAILCALL
+; CHECK-X86:        calll llvm.__logf128_finite.f128 at PLT
+start:
+  %0 = tail call fp128 @llvm.__logf128_finite.f128(fp128 %a)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.__logf128_finite.f128(fp128)
+
+
+define i64 @test_lrintf128(fp128 %a) {
+; CHECK-LABEL:      test_lrintf128:
+; CHECK-AARCH64:    b lrintl
+; CHECK-RISCV32:    call lrintl at plt
+; CHECK-S390X:      brasl {{%.*}} lrintl at PLT
+; CHECK-X64:        jmp lrintl at PLT
+; CHECK-X86:        calll lrintl
+start:
+  %0 = tail call i64 @llvm.lrint.f128(fp128 %a)
+  ret i64 %0
+}
+
+declare i64 @llvm.lrint.f128(fp128)
+
+
+define i64 @test_lroundf128(fp128 %a) {
+; CHECK-LABEL:      test_lroundf128:
+; CHECK-AARCH64:    b lroundl
+; CHECK-RISCV32:    call lroundl at plt
+; CHECK-S390X:      brasl {{%.*}} lroundl at PLT
+; CHECK-X64:        jmp lroundl at PLT
+; CHECK-X86:        calll lroundl
+start:
+  %0 = tail call i64 @llvm.lround.i64.f128(fp128 %a)
+  ret i64 %0
+}
+
+declare i64 @llvm.lround.i64.f128(fp128)
+
+
+define fp128 @test_nearbyintf128(fp128 %a) {
+; CHECK-LABEL:      test_nearbyintf128:
+; CHECK-AARCH64:    b nearbyintl
+; CHECK-RISCV32:    call nearbyintl at plt
+; CHECK-S390X:      brasl {{%.*}} nearbyintl at PLT
+; CHECK-X64:        jmp nearbyintl at PLT
+; CHECK-X86:        calll nearbyintl
+start:
+  %0 = tail call fp128 @llvm.nearbyint.f128(fp128 %a)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.nearbyint.f128(fp128)
+
+
+define fp128 @test_powf128(fp128 %a, fp128 %b) {
+; CHECK-LABEL:      test_powf128:
+; CHECK-AARCH64:    b powl
+; CHECK-RISCV32:    call powl at plt
+; CHECK-S390X:      brasl {{%.*}} powl at PLT
+; CHECK-X64:        jmp powl at PLT
+; CHECK-X86:        calll powl
+start:
+  %0 = tail call fp128 @llvm.pow.f128(fp128 %a, fp128 %b)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.pow.f128(fp128, fp128)
+
+
+define fp128 @test___powf128_finite(fp128 %a, fp128 %b) {
+; CHECK-LABEL:      test___powf128_finite:
+; CHECK-AARCH64:    b llvm.__powf128_finite.f128
+; CHECK-RISCV32:    call llvm.__powf128_finite.f128 at plt
+; CHECK-S390X:      brasl {{%.*}} llvm.__powf128_finite.f128 at PLT
+; CHECK-X64:        jmp llvm.__powf128_finite.f128 at PLT # TAILCALL
+; CHECK-X86:        calll llvm.__powf128_finite.f128 at PLT
+start:
+  %0 = tail call fp128 @llvm.__powf128_finite.f128(fp128 %a, fp128 %b)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.__powf128_finite.f128(fp128, fp128)
+
+
+define fp128 @test_rintf128(fp128 %a) {
+; CHECK-LABEL:      test_rintf128:
+; CHECK-AARCH64:    b rintl
+; CHECK-RISCV32:    call rintl at plt
+;
+; CHECK-S390X-LABEL: test_rintf128:
+; CHECK-S390X:       # %bb.0: # %start
+; CHECK-S390X-NEXT:    ld %f0, 0(%r3)
+; CHECK-S390X-NEXT:    ld %f2, 8(%r3)
+; CHECK-S390X-NEXT:    fixbr %f0, 0, %f0
+; CHECK-S390X-NEXT:    std %f0, 0(%r2)
+; CHECK-S390X-NEXT:    std %f2, 8(%r2)
+; CHECK-S390X-NEXT:    br %r14
+;
+; CHECK-X64:        jmp rintl at PLT
+; CHECK-X86:        calll rintl
+start:
+  %0 = tail call fp128 @llvm.rint.f128(fp128 %a)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.rint.f128(fp128)
+
+
+define fp128 @test_roundevenf128(fp128 %a) {
+; CHECK-LABEL:      test_roundevenf128:
+; CHECK-AARCH64:    b roundevenl
+; CHECK-RISCV32:    call roundevenl at plt
+; CHECK-S390X:      brasl {{%.*}} roundevenl at PLT
+; CHECK-X64:        jmp roundevenl at PLT
+; CHECK-X86:        calll roundevenl
+start:
+  %0 = tail call fp128 @llvm.roundeven.f128(fp128 %a)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.roundeven.f128(fp128)
+
+
+define fp128 @test_roundf128(fp128 %a) {
+; CHECK-LABEL:      test_roundf128:
+; CHECK-AARCH64:    b roundl
+; CHECK-RISCV32:    call roundl at plt
+; CHECK-S390X:      brasl {{%.*}} roundl at PLT
+; CHECK-X64:        jmp roundl at PLT
+; CHECK-X86:        calll roundl
+start:
+  %0 = tail call fp128 @llvm.round.f128(fp128 %a)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.round.f128(fp128)
+
+
+define fp128 @test_sinf128(fp128 %a) {
+; CHECK-LABEL:      test_sinf128:
+; CHECK-AARCH64:    b sinl
+; CHECK-RISCV32:    call sinl at plt
+; CHECK-S390X:      brasl {{%.*}} sinl at PLT
+; CHECK-X64:        jmp sinl at PLT
+; CHECK-X86:        calll sinl
+start:
+  %0 = tail call fp128 @llvm.sin.f128(fp128 %a)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.sin.f128(fp128)
+
+
+define fp128 @test_sqrtf128(fp128 %a) {
+; CHECK-LABEL:      test_sqrtf128:
+; CHECK-AARCH64:    b sqrtl
+; CHECK-RISCV32:    call sqrtl at plt
+; CHECK-S390X:      sqxbr {{%.*}} {{%.*}}
+; CHECK-X64:        jmp sqrtl at PLT
+; CHECK-X86:        calll sqrtl
+start:
+  %0 = tail call fp128 @llvm.sqrt.f128(fp128 %a)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.sqrt.f128(fp128)
+
+
+define fp128 @test_truncf128(fp128 %a) {
+; CHECK-LABEL:      test_truncf128:
+; CHECK-AARCH64:    b truncl
+; CHECK-RISCV32:    call truncl at plt
+; CHECK-S390X:      brasl {{%.*}} truncl at PLT
+; CHECK-X64:        jmp truncl at PLT
+; CHECK-X86:        calll truncl
+start:
+  %0 = tail call fp128 @llvm.trunc.f128(fp128 %a)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.trunc.f128(fp128)

>From 76e1a7712d7c4964221c7b293ad4888ba91ac3d7 Mon Sep 17 00:00:00 2001
From: Trevor Gross <tmgross at umich.edu>
Date: Thu, 28 Dec 2023 03:50:05 -0500
Subject: [PATCH 2/4] [llvm][clang] Move `long double` layout logic from clang
 to LLVM

Information about the size and alignment of `long double` is currently part of
clang. Move this logic to LLVM so it can be used to control lowering of
intrinsics.
---
 clang/include/clang/Basic/TargetInfo.h  |   6 ++
 clang/lib/Basic/TargetInfo.cpp          |  10 +-
 clang/lib/Basic/Targets/AArch64.cpp     |   8 +-
 clang/lib/Basic/Targets/ARC.h           |   2 +-
 clang/lib/Basic/Targets/AVR.h           |   3 -
 clang/lib/Basic/Targets/CSKY.h          |   2 +-
 clang/lib/Basic/Targets/LoongArch.h     |   3 -
 clang/lib/Basic/Targets/MSP430.h        |   4 +-
 clang/lib/Basic/Targets/Mips.h          |   8 --
 clang/lib/Basic/Targets/RISCV.h         |   3 -
 clang/lib/Basic/Targets/Sparc.h         |   6 +-
 clang/lib/Basic/Targets/SystemZ.h       |   3 -
 clang/lib/Basic/Targets/TCE.h           |   3 -
 clang/lib/Basic/Targets/VE.h            |   3 -
 clang/lib/Basic/Targets/WebAssembly.h   |   2 -
 clang/lib/Basic/Targets/X86.h           |  11 +--
 clang/lib/Basic/Targets/XCore.h         |   2 +-
 llvm/include/llvm/TargetParser/Triple.h |  18 ++++
 llvm/lib/TargetParser/Triple.cpp        | 118 ++++++++++++++++++++++++
 19 files changed, 158 insertions(+), 57 deletions(-)

diff --git a/clang/include/clang/Basic/TargetInfo.h b/clang/include/clang/Basic/TargetInfo.h
index ac3c324c6c29c4..f6e6d24acaed1b 100644
--- a/clang/include/clang/Basic/TargetInfo.h
+++ b/clang/include/clang/Basic/TargetInfo.h
@@ -1218,6 +1218,12 @@ class TargetInfo : public TransferrableTargetInfo,
     return Triple;
   }
 
+  /// Returns information about C layouts that come from LLVM. This is used
+  /// to control lowering to libcalls.
+  llvm::Triple::CLayouts getTripleLayouts() const {
+    return Triple.getCLayouts();
+  }
+
   /// Returns the target ID if supported.
   virtual std::optional<std::string> getTargetID() const {
     return std::nullopt;
diff --git a/clang/lib/Basic/TargetInfo.cpp b/clang/lib/Basic/TargetInfo.cpp
index 96b3ad9ba2f273..2cf4cacf490cf5 100644
--- a/clang/lib/Basic/TargetInfo.cpp
+++ b/clang/lib/Basic/TargetInfo.cpp
@@ -54,6 +54,10 @@ static const LangASMap FakeAddrSpaceMap = {
 TargetInfo::TargetInfo(const llvm::Triple &T) : Triple(T) {
   // Set defaults.  Defaults are set for a 32-bit RISC platform, like PPC or
   // SPARC.  These should be overridden by concrete targets as needed.
+
+  // Load some defaults from LLVM
+  llvm::Triple::CLayouts TripleLayouts = getTripleLayouts();
+
   BigEndian = !T.isLittleEndian();
   TLSSupported = true;
   VLASupported = true;
@@ -113,8 +117,8 @@ TargetInfo::TargetInfo(const llvm::Triple &T) : Triple(T) {
   FloatAlign = 32;
   DoubleWidth = 64;
   DoubleAlign = 64;
-  LongDoubleWidth = 64;
-  LongDoubleAlign = 64;
+  LongDoubleWidth = TripleLayouts.LongDoubleWidth;
+  LongDoubleAlign = TripleLayouts.LongDoubleAlign;
   Float128Align = 128;
   Ibm128Align = 128;
   LargeArrayMinWidth = 0;
@@ -144,7 +148,7 @@ TargetInfo::TargetInfo(const llvm::Triple &T) : Triple(T) {
   HalfFormat = &llvm::APFloat::IEEEhalf();
   FloatFormat = &llvm::APFloat::IEEEsingle();
   DoubleFormat = &llvm::APFloat::IEEEdouble();
-  LongDoubleFormat = &llvm::APFloat::IEEEdouble();
+  LongDoubleFormat = TripleLayouts.LongDoubleFormat;
   Float128Format = &llvm::APFloat::IEEEquad();
   Ibm128Format = &llvm::APFloat::PPCDoubleDouble();
   MCountName = "mcount";
diff --git a/clang/lib/Basic/Targets/AArch64.cpp b/clang/lib/Basic/Targets/AArch64.cpp
index 2f8395cb8932f2..c3ed06ca550b61 100644
--- a/clang/lib/Basic/Targets/AArch64.cpp
+++ b/clang/lib/Basic/Targets/AArch64.cpp
@@ -156,8 +156,7 @@ AArch64TargetInfo::AArch64TargetInfo(const llvm::Triple &Triple,
   MaxAtomicInlineWidth = 128;
   MaxAtomicPromoteWidth = 128;
 
-  LongDoubleWidth = LongDoubleAlign = SuitableAlign = 128;
-  LongDoubleFormat = &llvm::APFloat::IEEEquad();
+  SuitableAlign = LongDoubleAlign;
 
   BFloat16Width = BFloat16Align = 16;
   BFloat16Format = &llvm::APFloat::BFloat();
@@ -1452,8 +1451,6 @@ WindowsARM64TargetInfo::WindowsARM64TargetInfo(const llvm::Triple &Triple,
   IntWidth = IntAlign = 32;
   LongWidth = LongAlign = 32;
   DoubleAlign = LongLongAlign = 64;
-  LongDoubleWidth = LongDoubleAlign = 64;
-  LongDoubleFormat = &llvm::APFloat::IEEEdouble();
   IntMaxType = SignedLongLong;
   Int64Type = SignedLongLong;
   SizeType = UnsignedLongLong;
@@ -1550,8 +1547,7 @@ DarwinAArch64TargetInfo::DarwinAArch64TargetInfo(const llvm::Triple &Triple,
   WCharType = SignedInt;
   UseSignedCharForObjCBool = false;
 
-  LongDoubleWidth = LongDoubleAlign = SuitableAlign = 64;
-  LongDoubleFormat = &llvm::APFloat::IEEEdouble();
+  SuitableAlign = LongDoubleAlign;
 
   UseZeroLengthBitfieldAlignment = false;
 
diff --git a/clang/lib/Basic/Targets/ARC.h b/clang/lib/Basic/Targets/ARC.h
index fcbfdd6eec5862..b9da57e6e16448 100644
--- a/clang/lib/Basic/Targets/ARC.h
+++ b/clang/lib/Basic/Targets/ARC.h
@@ -28,7 +28,7 @@ class LLVM_LIBRARY_VISIBILITY ARCTargetInfo : public TargetInfo {
     NoAsmVariants = true;
     LongLongAlign = 32;
     SuitableAlign = 32;
-    DoubleAlign = LongDoubleAlign = 32;
+    DoubleAlign = 32;
     SizeType = UnsignedInt;
     PtrDiffType = SignedInt;
     IntPtrType = SignedInt;
diff --git a/clang/lib/Basic/Targets/AVR.h b/clang/lib/Basic/Targets/AVR.h
index 854a51d78c393b..38847b057991f3 100644
--- a/clang/lib/Basic/Targets/AVR.h
+++ b/clang/lib/Basic/Targets/AVR.h
@@ -44,9 +44,6 @@ class LLVM_LIBRARY_VISIBILITY AVRTargetInfo : public TargetInfo {
     DoubleWidth = 32;
     DoubleAlign = 8;
     DoubleFormat = &llvm::APFloat::IEEEsingle();
-    LongDoubleWidth = 32;
-    LongDoubleAlign = 8;
-    LongDoubleFormat = &llvm::APFloat::IEEEsingle();
     SizeType = UnsignedInt;
     PtrDiffType = SignedInt;
     IntPtrType = SignedInt;
diff --git a/clang/lib/Basic/Targets/CSKY.h b/clang/lib/Basic/Targets/CSKY.h
index 11404e37db368a..c600c0fe02152c 100644
--- a/clang/lib/Basic/Targets/CSKY.h
+++ b/clang/lib/Basic/Targets/CSKY.h
@@ -43,7 +43,7 @@ class LLVM_LIBRARY_VISIBILITY CSKYTargetInfo : public TargetInfo {
     NoAsmVariants = true;
     LongLongAlign = 32;
     SuitableAlign = 32;
-    DoubleAlign = LongDoubleAlign = 32;
+    DoubleAlign = 32;
     SizeType = UnsignedInt;
     PtrDiffType = SignedInt;
     IntPtrType = SignedInt;
diff --git a/clang/lib/Basic/Targets/LoongArch.h b/clang/lib/Basic/Targets/LoongArch.h
index 3313102492cb8d..894d7ce9fa7781 100644
--- a/clang/lib/Basic/Targets/LoongArch.h
+++ b/clang/lib/Basic/Targets/LoongArch.h
@@ -37,9 +37,6 @@ class LLVM_LIBRARY_VISIBILITY LoongArchTargetInfo : public TargetInfo {
     HasFeatureF = false;
     HasFeatureLSX = false;
     HasFeatureLASX = false;
-    LongDoubleWidth = 128;
-    LongDoubleAlign = 128;
-    LongDoubleFormat = &llvm::APFloat::IEEEquad();
     MCountName = "_mcount";
     SuitableAlign = 128;
     WCharType = SignedInt;
diff --git a/clang/lib/Basic/Targets/MSP430.h b/clang/lib/Basic/Targets/MSP430.h
index 25639b8c1e0ad9..d692398a80aee8 100644
--- a/clang/lib/Basic/Targets/MSP430.h
+++ b/clang/lib/Basic/Targets/MSP430.h
@@ -35,8 +35,8 @@ class LLVM_LIBRARY_VISIBILITY MSP430TargetInfo : public TargetInfo {
     LongAlign = LongLongAlign = 16;
     FloatWidth = 32;
     FloatAlign = 16;
-    DoubleWidth = LongDoubleWidth = 64;
-    DoubleAlign = LongDoubleAlign = 16;
+    DoubleWidth = 64;
+    DoubleAlign = 16;
     PointerWidth = 16;
     PointerAlign = 16;
     SuitableAlign = 16;
diff --git a/clang/lib/Basic/Targets/Mips.h b/clang/lib/Basic/Targets/Mips.h
index f46b95abfd75c7..3387adfd51ddae 100644
--- a/clang/lib/Basic/Targets/Mips.h
+++ b/clang/lib/Basic/Targets/Mips.h
@@ -118,8 +118,6 @@ class LLVM_LIBRARY_VISIBILITY MipsTargetInfo : public TargetInfo {
   void setO32ABITypes() {
     Int64Type = SignedLongLong;
     IntMaxType = Int64Type;
-    LongDoubleFormat = &llvm::APFloat::IEEEdouble();
-    LongDoubleWidth = LongDoubleAlign = 64;
     LongWidth = LongAlign = 32;
     MaxAtomicPromoteWidth = MaxAtomicInlineWidth = 32;
     PointerWidth = PointerAlign = 32;
@@ -129,12 +127,6 @@ class LLVM_LIBRARY_VISIBILITY MipsTargetInfo : public TargetInfo {
   }
 
   void setN32N64ABITypes() {
-    LongDoubleWidth = LongDoubleAlign = 128;
-    LongDoubleFormat = &llvm::APFloat::IEEEquad();
-    if (getTriple().isOSFreeBSD()) {
-      LongDoubleWidth = LongDoubleAlign = 64;
-      LongDoubleFormat = &llvm::APFloat::IEEEdouble();
-    }
     MaxAtomicPromoteWidth = MaxAtomicInlineWidth = 64;
     SuitableAlign = 128;
   }
diff --git a/clang/lib/Basic/Targets/RISCV.h b/clang/lib/Basic/Targets/RISCV.h
index f98c88cd45f831..53a35c1cb75b99 100644
--- a/clang/lib/Basic/Targets/RISCV.h
+++ b/clang/lib/Basic/Targets/RISCV.h
@@ -39,9 +39,6 @@ class RISCVTargetInfo : public TargetInfo {
     BFloat16Width = 16;
     BFloat16Align = 16;
     BFloat16Format = &llvm::APFloat::BFloat();
-    LongDoubleWidth = 128;
-    LongDoubleAlign = 128;
-    LongDoubleFormat = &llvm::APFloat::IEEEquad();
     SuitableAlign = 128;
     WCharType = SignedInt;
     WIntType = UnsignedInt;
diff --git a/clang/lib/Basic/Targets/Sparc.h b/clang/lib/Basic/Targets/Sparc.h
index 214fef88e1dcd4..a226932421ac50 100644
--- a/clang/lib/Basic/Targets/Sparc.h
+++ b/clang/lib/Basic/Targets/Sparc.h
@@ -205,12 +205,8 @@ class LLVM_LIBRARY_VISIBILITY SparcV9TargetInfo : public SparcTargetInfo {
       IntMaxType = SignedLong;
     Int64Type = IntMaxType;
 
-    // The SPARCv8 System V ABI has long double 128-bits in size, but 64-bit
-    // aligned. The SPARCv9 SCD 2.4.1 says 16-byte aligned.
-    LongDoubleWidth = 128;
-    LongDoubleAlign = 128;
     SuitableAlign = 128;
-    LongDoubleFormat = &llvm::APFloat::IEEEquad();
+
     MaxAtomicPromoteWidth = MaxAtomicInlineWidth = 64;
   }
 
diff --git a/clang/lib/Basic/Targets/SystemZ.h b/clang/lib/Basic/Targets/SystemZ.h
index e4ec338880f210..e244fd0110ab67 100644
--- a/clang/lib/Basic/Targets/SystemZ.h
+++ b/clang/lib/Basic/Targets/SystemZ.h
@@ -40,9 +40,6 @@ class LLVM_LIBRARY_VISIBILITY SystemZTargetInfo : public TargetInfo {
     LongWidth = LongLongWidth = LongAlign = LongLongAlign = 64;
     Int128Align = 64;
     PointerWidth = PointerAlign = 64;
-    LongDoubleWidth = 128;
-    LongDoubleAlign = 64;
-    LongDoubleFormat = &llvm::APFloat::IEEEquad();
     DefaultAlignForAttributeAligned = 64;
     MinGlobalAlign = 16;
     if (Triple.isOSzOS()) {
diff --git a/clang/lib/Basic/Targets/TCE.h b/clang/lib/Basic/Targets/TCE.h
index dcf684fe6dbc01..71ec11a7724f5d 100644
--- a/clang/lib/Basic/Targets/TCE.h
+++ b/clang/lib/Basic/Targets/TCE.h
@@ -76,11 +76,8 @@ class LLVM_LIBRARY_VISIBILITY TCETargetInfo : public TargetInfo {
     FloatAlign = 32;
     DoubleWidth = 32;
     DoubleAlign = 32;
-    LongDoubleWidth = 32;
-    LongDoubleAlign = 32;
     FloatFormat = &llvm::APFloat::IEEEsingle();
     DoubleFormat = &llvm::APFloat::IEEEsingle();
-    LongDoubleFormat = &llvm::APFloat::IEEEsingle();
     resetDataLayout("E-p:32:32:32-i1:8:8-i8:8:32-"
                     "i16:16:32-i32:32:32-i64:32:32-"
                     "f32:32:32-f64:32:32-v64:32:32-"
diff --git a/clang/lib/Basic/Targets/VE.h b/clang/lib/Basic/Targets/VE.h
index ea9a092cad8090..09ca50b9dea0d1 100644
--- a/clang/lib/Basic/Targets/VE.h
+++ b/clang/lib/Basic/Targets/VE.h
@@ -27,9 +27,6 @@ class LLVM_LIBRARY_VISIBILITY VETargetInfo : public TargetInfo {
   VETargetInfo(const llvm::Triple &Triple, const TargetOptions &)
       : TargetInfo(Triple) {
     NoAsmVariants = true;
-    LongDoubleWidth = 128;
-    LongDoubleAlign = 128;
-    LongDoubleFormat = &llvm::APFloat::IEEEquad();
     DoubleAlign = LongLongAlign = 64;
     SuitableAlign = 64;
     LongWidth = LongAlign = PointerWidth = PointerAlign = 64;
diff --git a/clang/lib/Basic/Targets/WebAssembly.h b/clang/lib/Basic/Targets/WebAssembly.h
index 83b1711f9fdf6a..916692786722cd 100644
--- a/clang/lib/Basic/Targets/WebAssembly.h
+++ b/clang/lib/Basic/Targets/WebAssembly.h
@@ -76,8 +76,6 @@ class LLVM_LIBRARY_VISIBILITY WebAssemblyTargetInfo : public TargetInfo {
     LargeArrayMinWidth = 128;
     LargeArrayAlign = 128;
     SigAtomicType = SignedLong;
-    LongDoubleWidth = LongDoubleAlign = 128;
-    LongDoubleFormat = &llvm::APFloat::IEEEquad();
     MaxAtomicPromoteWidth = MaxAtomicInlineWidth = 64;
     // size_t being unsigned long for both wasm32 and wasm64 makes mangled names
     // more consistent between the two.
diff --git a/clang/lib/Basic/Targets/X86.h b/clang/lib/Basic/Targets/X86.h
index 0ab1c10833db26..f944726587beda 100644
--- a/clang/lib/Basic/Targets/X86.h
+++ b/clang/lib/Basic/Targets/X86.h
@@ -185,7 +185,6 @@ class LLVM_LIBRARY_VISIBILITY X86TargetInfo : public TargetInfo {
       : TargetInfo(Triple) {
     BFloat16Width = BFloat16Align = 16;
     BFloat16Format = &llvm::APFloat::BFloat();
-    LongDoubleFormat = &llvm::APFloat::x87DoubleExtended();
     AddrSpaceMap = &X86AddrSpaceMap;
     HasStrictFP = true;
 
@@ -437,8 +436,6 @@ class LLVM_LIBRARY_VISIBILITY X86_32TargetInfo : public X86TargetInfo {
   X86_32TargetInfo(const llvm::Triple &Triple, const TargetOptions &Opts)
       : X86TargetInfo(Triple, Opts) {
     DoubleAlign = LongLongAlign = 32;
-    LongDoubleWidth = 96;
-    LongDoubleAlign = 32;
     SuitableAlign = 128;
     resetDataLayout(Triple.isOSBinFormatMachO()
                         ? "e-m:o-p:32:32-p270:32:32-p271:32:32-p272:64:64-i128:"
@@ -540,8 +537,6 @@ class LLVM_LIBRARY_VISIBILITY DarwinI386TargetInfo
 public:
   DarwinI386TargetInfo(const llvm::Triple &Triple, const TargetOptions &Opts)
       : DarwinTargetInfo<X86_32TargetInfo>(Triple, Opts) {
-    LongDoubleWidth = 128;
-    LongDoubleAlign = 128;
     SuitableAlign = 128;
     MaxVectorAlign = 256;
     // The watchOS simulator uses the builtin bool type for Objective-C.
@@ -987,8 +982,6 @@ class LLVM_LIBRARY_VISIBILITY OHOSX86_32TargetInfo
   OHOSX86_32TargetInfo(const llvm::Triple &Triple, const TargetOptions &Opts)
       : OHOSTargetInfo<X86_32TargetInfo>(Triple, Opts) {
     SuitableAlign = 32;
-    LongDoubleWidth = 64;
-    LongDoubleFormat = &llvm::APFloat::IEEEdouble();
   }
 };
 
@@ -997,9 +990,7 @@ class LLVM_LIBRARY_VISIBILITY OHOSX86_64TargetInfo
     : public OHOSTargetInfo<X86_64TargetInfo> {
 public:
   OHOSX86_64TargetInfo(const llvm::Triple &Triple, const TargetOptions &Opts)
-      : OHOSTargetInfo<X86_64TargetInfo>(Triple, Opts) {
-    LongDoubleFormat = &llvm::APFloat::IEEEquad();
-  }
+      : OHOSTargetInfo<X86_64TargetInfo>(Triple, Opts) {}
 };
 } // namespace targets
 } // namespace clang
diff --git a/clang/lib/Basic/Targets/XCore.h b/clang/lib/Basic/Targets/XCore.h
index a58d3e8acf4791..e96584d6c1db5a 100644
--- a/clang/lib/Basic/Targets/XCore.h
+++ b/clang/lib/Basic/Targets/XCore.h
@@ -29,7 +29,7 @@ class LLVM_LIBRARY_VISIBILITY XCoreTargetInfo : public TargetInfo {
     NoAsmVariants = true;
     LongLongAlign = 32;
     SuitableAlign = 32;
-    DoubleAlign = LongDoubleAlign = 32;
+    DoubleAlign = 32;
     SizeType = UnsignedInt;
     PtrDiffType = SignedInt;
     IntPtrType = SignedInt;
diff --git a/llvm/include/llvm/TargetParser/Triple.h b/llvm/include/llvm/TargetParser/Triple.h
index 47904621c0967f..8bc9f03b789f84 100644
--- a/llvm/include/llvm/TargetParser/Triple.h
+++ b/llvm/include/llvm/TargetParser/Triple.h
@@ -9,6 +9,7 @@
 #ifndef LLVM_TARGETPARSER_TRIPLE_H
 #define LLVM_TARGETPARSER_TRIPLE_H
 
+#include "llvm/ADT/APFloat.h"
 #include "llvm/ADT/Twine.h"
 #include "llvm/Support/VersionTuple.h"
 
@@ -1155,6 +1156,23 @@ class Triple {
   /// Returns a canonicalized OS version number for the specified OS.
   static VersionTuple getCanonicalVersionForOS(OSType OSKind,
                                                const VersionTuple &Version);
+
+  /// Layouts for C types that are relevant to libc calls generated by LLVM
+  struct CLayouts {
+    unsigned char LongDoubleWidth;
+    unsigned char LongDoubleAlign;
+    const fltSemantics *LongDoubleFormat;
+  };
+
+  /// Provide default layouts relevant to C. Frontends may override these
+  /// values.
+  CLayouts getCLayouts() const;
+
+  /// Return true if `long double` and `__float128` have the same layout.
+  bool isLongDoubleF128() const {
+    // TODO: do we also need to check alignment?
+    return getCLayouts().LongDoubleWidth == 128;
+  }
 };
 
 } // End llvm namespace
diff --git a/llvm/lib/TargetParser/Triple.cpp b/llvm/lib/TargetParser/Triple.cpp
index e93502187b5492..a2f3cb8db39e03 100644
--- a/llvm/lib/TargetParser/Triple.cpp
+++ b/llvm/lib/TargetParser/Triple.cpp
@@ -1897,6 +1897,124 @@ VersionTuple Triple::getCanonicalVersionForOS(OSType OSKind,
   }
 }
 
+Triple::CLayouts Triple::getCLayouts() const {
+  Triple::CLayouts Layouts;
+
+  // Default to a 32-bit RISC platform
+  Layouts.LongDoubleWidth = 64;
+  Layouts.LongDoubleAlign = 64;
+  Layouts.LongDoubleFormat = &llvm::APFloat::IEEEdouble();
+
+  enum ArchType arch = getArch();
+
+  if (arch == aarch64 || arch == aarch64_be || arch == aarch64_32) {
+    Layouts.LongDoubleWidth = 128;
+    Layouts.LongDoubleAlign = 128;
+    Layouts.LongDoubleFormat = &llvm::APFloat::IEEEquad();
+
+    // TODO: verify this logic matches when WindowsARM64TargetInfo /
+    // DarwinAArch64TargetInfo is called
+    if (isOSWindows()) {
+      Layouts.LongDoubleWidth = Layouts.LongDoubleAlign = 64;
+      Layouts.LongDoubleFormat = &llvm::APFloat::IEEEdouble();
+    } else if (isMacOSX()) {
+      // TODO: should this just be isMacOSX or check specifically for darwin?
+      Layouts.LongDoubleWidth = Layouts.LongDoubleAlign = 64;
+      Layouts.LongDoubleFormat = &llvm::APFloat::IEEEdouble();
+    }
+  } else if (arch == avr) {
+    Layouts.LongDoubleWidth = 32;
+    Layouts.LongDoubleAlign = 8;
+    Layouts.LongDoubleFormat = &llvm::APFloat::IEEEsingle();
+  } else if (arch == arc) {
+    Layouts.LongDoubleAlign = 32;
+  } else if (arch == arm) {
+    // TODO: port the logic
+  } else if (arch == csky) {
+    Layouts.LongDoubleAlign = 32;
+  } else if (arch == loongarch32 || arch == loongarch64) {
+    Layouts.LongDoubleWidth = Layouts.LongDoubleAlign = 128;
+    Layouts.LongDoubleFormat = &llvm::APFloat::IEEEquad();
+  } else if (arch == mips || arch == mipsel || arch == mips64 ||
+             arch == mips64el) {
+    if (isMIPS32()) {
+      // o32
+      Layouts.LongDoubleWidth = Layouts.LongDoubleAlign = 64;
+      Layouts.LongDoubleFormat = &llvm::APFloat::IEEEdouble();
+    } else {
+      // n32 & n64
+      Layouts.LongDoubleWidth = Layouts.LongDoubleAlign = 128;
+      Layouts.LongDoubleFormat = &llvm::APFloat::IEEEquad();
+      if (isOSFreeBSD()) {
+        Layouts.LongDoubleWidth = Layouts.LongDoubleAlign = 64;
+        Layouts.LongDoubleFormat = &llvm::APFloat::IEEEdouble();
+      }
+    }
+  } else if (arch == msp430) {
+    Layouts.LongDoubleWidth = 64;
+    Layouts.LongDoubleAlign = 16;
+  } else if (arch == ppc || arch == ppcle || arch == ppc64 || arch == ppc64le) {
+    // TODO: figure out how to get features
+  } else if (arch == riscv32 || arch == riscv64) {
+    Layouts.LongDoubleWidth = 128;
+    Layouts.LongDoubleAlign = 128;
+    Layouts.LongDoubleFormat = &llvm::APFloat::IEEEquad();
+  } else if (arch == sparcv9) {
+    // The SPARCv8 System V ABI has long double 128-bits in size, but 64-bit
+    // aligned. The SPARCv9 SCD 2.4.1 says 16-byte aligned.
+    Layouts.LongDoubleWidth = 128;
+    Layouts.LongDoubleAlign = 128;
+    Layouts.LongDoubleFormat = &llvm::APFloat::IEEEquad();
+  } else if (arch == systemz) {
+    Layouts.LongDoubleWidth = 128;
+    Layouts.LongDoubleAlign = 64;
+    Layouts.LongDoubleFormat = &llvm::APFloat::IEEEquad();
+  } else if (arch == tce || arch == tcele) {
+    Layouts.LongDoubleWidth = 32;
+    Layouts.LongDoubleAlign = 32;
+    Layouts.LongDoubleFormat = &llvm::APFloat::IEEEsingle();
+  } else if (arch == ve) {
+    Layouts.LongDoubleWidth = 128;
+    Layouts.LongDoubleAlign = 128;
+    Layouts.LongDoubleFormat = &llvm::APFloat::IEEEquad();
+  } else if (arch == wasm32 || arch == wasm64) {
+    Layouts.LongDoubleWidth = Layouts.LongDoubleAlign = 128;
+    Layouts.LongDoubleFormat = &llvm::APFloat::IEEEquad();
+  } else if (arch == x86) {
+    Layouts.LongDoubleWidth = 96;
+    Layouts.LongDoubleAlign = 32;
+    Layouts.LongDoubleFormat = &llvm::APFloat::x87DoubleExtended();
+    if (isMacOSX()) {
+      Layouts.LongDoubleWidth = 128;
+      Layouts.LongDoubleAlign = 128;
+    } else if (isOSWindows()) {
+      // TODO: how to get difference between WindowsX86_32TargetInfo,
+      // MicrosoftX86_32TargetInfo, CygwinX86_64TargetInfo, and
+      // MinGWX86_32TargetInfo
+      Layouts.LongDoubleWidth = Layouts.LongDoubleAlign = 64;
+      Layouts.LongDoubleFormat = &llvm::APFloat::IEEEdouble();
+    } else if (isOSIAMCU()) {
+      Layouts.LongDoubleWidth = 64;
+      Layouts.LongDoubleFormat = &llvm::APFloat::IEEEdouble();
+    } else if (isOHOSFamily()) {
+      Layouts.LongDoubleWidth = 64;
+      Layouts.LongDoubleFormat = &llvm::APFloat::IEEEdouble();
+    }
+  } else if (arch == x86_64) {
+    Layouts.LongDoubleWidth = 128;
+    Layouts.LongDoubleAlign = 128;
+    Layouts.LongDoubleFormat = &llvm::APFloat::x87DoubleExtended();
+
+    if (isOHOSFamily()) {
+      Layouts.LongDoubleFormat = &llvm::APFloat::IEEEquad();
+    }
+  } else if (arch == xcore) {
+    Layouts.LongDoubleAlign = 32;
+  }
+
+  return Layouts;
+}
+
 // HLSL triple environment orders are relied on in the front end
 static_assert(Triple::Vertex - Triple::Pixel == 1,
               "incorrect HLSL stage order");

>From ef74f65868e727261ddd604e86cf78c715b3a416 Mon Sep 17 00:00:00 2001
From: Trevor Gross <tmgross at umich.edu>
Date: Thu, 28 Dec 2023 04:01:22 -0500
Subject: [PATCH 3/4] [IR] Change `fp128` lowering to use `f128` functions by
 default

Switch from emitting long double functions to using `f128`-specific functions.

Fixes https://github.com/llvm/llvm-project/issues/44744.
---
 llvm/include/llvm/IR/RuntimeLibcalls.def      |  68 +--
 .../CodeGen/Generic/f128-math-lowering.ll     | 226 ++++----
 llvm/test/CodeGen/X86/f128-arith.ll           | 515 ++++++++++++++++++
 3 files changed, 662 insertions(+), 147 deletions(-)
 create mode 100644 llvm/test/CodeGen/X86/f128-arith.ll

diff --git a/llvm/include/llvm/IR/RuntimeLibcalls.def b/llvm/include/llvm/IR/RuntimeLibcalls.def
index 19dea60bebf9be..7fce540612856a 100644
--- a/llvm/include/llvm/IR/RuntimeLibcalls.def
+++ b/llvm/include/llvm/IR/RuntimeLibcalls.def
@@ -110,12 +110,12 @@ HANDLE_LIBCALL(DIV_PPCF128, "__gcc_qdiv")
 HANDLE_LIBCALL(REM_F32, "fmodf")
 HANDLE_LIBCALL(REM_F64, "fmod")
 HANDLE_LIBCALL(REM_F80, "fmodl")
-HANDLE_LIBCALL(REM_F128, "fmodl")
+HANDLE_LIBCALL(REM_F128, "fmodf128")
 HANDLE_LIBCALL(REM_PPCF128, "fmodl")
 HANDLE_LIBCALL(FMA_F32, "fmaf")
 HANDLE_LIBCALL(FMA_F64, "fma")
 HANDLE_LIBCALL(FMA_F80, "fmal")
-HANDLE_LIBCALL(FMA_F128, "fmal")
+HANDLE_LIBCALL(FMA_F128, "fmaf128")
 HANDLE_LIBCALL(FMA_PPCF128, "fmal")
 HANDLE_LIBCALL(POWI_F32, "__powisf2")
 HANDLE_LIBCALL(POWI_F64, "__powidf2")
@@ -125,62 +125,62 @@ HANDLE_LIBCALL(POWI_PPCF128, "__powitf2")
 HANDLE_LIBCALL(SQRT_F32, "sqrtf")
 HANDLE_LIBCALL(SQRT_F64, "sqrt")
 HANDLE_LIBCALL(SQRT_F80, "sqrtl")
-HANDLE_LIBCALL(SQRT_F128, "sqrtl")
+HANDLE_LIBCALL(SQRT_F128, "sqrtf128")
 HANDLE_LIBCALL(SQRT_PPCF128, "sqrtl")
 HANDLE_LIBCALL(CBRT_F32, "cbrtf")
 HANDLE_LIBCALL(CBRT_F64, "cbrt")
 HANDLE_LIBCALL(CBRT_F80, "cbrtl")
-HANDLE_LIBCALL(CBRT_F128, "cbrtl")
+HANDLE_LIBCALL(CBRT_F128, "cbrtf128")
 HANDLE_LIBCALL(CBRT_PPCF128, "cbrtl")
 HANDLE_LIBCALL(LOG_F32, "logf")
 HANDLE_LIBCALL(LOG_F64, "log")
 HANDLE_LIBCALL(LOG_F80, "logl")
-HANDLE_LIBCALL(LOG_F128, "logl")
+HANDLE_LIBCALL(LOG_F128, "logf128")
 HANDLE_LIBCALL(LOG_PPCF128, "logl")
 HANDLE_LIBCALL(LOG_FINITE_F32, "__logf_finite")
 HANDLE_LIBCALL(LOG_FINITE_F64, "__log_finite")
 HANDLE_LIBCALL(LOG_FINITE_F80, "__logl_finite")
-HANDLE_LIBCALL(LOG_FINITE_F128, "__logl_finite")
+HANDLE_LIBCALL(LOG_FINITE_F128, "__logf128_finite")
 HANDLE_LIBCALL(LOG_FINITE_PPCF128, "__logl_finite")
 HANDLE_LIBCALL(LOG2_F32, "log2f")
 HANDLE_LIBCALL(LOG2_F64, "log2")
 HANDLE_LIBCALL(LOG2_F80, "log2l")
-HANDLE_LIBCALL(LOG2_F128, "log2l")
+HANDLE_LIBCALL(LOG2_F128, "log2f128")
 HANDLE_LIBCALL(LOG2_PPCF128, "log2l")
 HANDLE_LIBCALL(LOG2_FINITE_F32, "__log2f_finite")
 HANDLE_LIBCALL(LOG2_FINITE_F64, "__log2_finite")
 HANDLE_LIBCALL(LOG2_FINITE_F80, "__log2l_finite")
-HANDLE_LIBCALL(LOG2_FINITE_F128, "__log2l_finite")
+HANDLE_LIBCALL(LOG2_FINITE_F128, "__log2f128_finite")
 HANDLE_LIBCALL(LOG2_FINITE_PPCF128, "__log2l_finite")
 HANDLE_LIBCALL(LOG10_F32, "log10f")
 HANDLE_LIBCALL(LOG10_F64, "log10")
 HANDLE_LIBCALL(LOG10_F80, "log10l")
-HANDLE_LIBCALL(LOG10_F128, "log10l")
+HANDLE_LIBCALL(LOG10_F128, "log10f128")
 HANDLE_LIBCALL(LOG10_PPCF128, "log10l")
 HANDLE_LIBCALL(LOG10_FINITE_F32, "__log10f_finite")
 HANDLE_LIBCALL(LOG10_FINITE_F64, "__log10_finite")
 HANDLE_LIBCALL(LOG10_FINITE_F80, "__log10l_finite")
-HANDLE_LIBCALL(LOG10_FINITE_F128, "__log10l_finite")
+HANDLE_LIBCALL(LOG10_FINITE_F128, "__log10f128_finite")
 HANDLE_LIBCALL(LOG10_FINITE_PPCF128, "__log10l_finite")
 HANDLE_LIBCALL(EXP_F32, "expf")
 HANDLE_LIBCALL(EXP_F64, "exp")
 HANDLE_LIBCALL(EXP_F80, "expl")
-HANDLE_LIBCALL(EXP_F128, "expl")
+HANDLE_LIBCALL(EXP_F128, "expf128")
 HANDLE_LIBCALL(EXP_PPCF128, "expl")
 HANDLE_LIBCALL(EXP_FINITE_F32, "__expf_finite")
 HANDLE_LIBCALL(EXP_FINITE_F64, "__exp_finite")
 HANDLE_LIBCALL(EXP_FINITE_F80, "__expl_finite")
-HANDLE_LIBCALL(EXP_FINITE_F128, "__expl_finite")
+HANDLE_LIBCALL(EXP_FINITE_F128, "__expf128_finite")
 HANDLE_LIBCALL(EXP_FINITE_PPCF128, "__expl_finite")
 HANDLE_LIBCALL(EXP2_F32, "exp2f")
 HANDLE_LIBCALL(EXP2_F64, "exp2")
 HANDLE_LIBCALL(EXP2_F80, "exp2l")
-HANDLE_LIBCALL(EXP2_F128, "exp2l")
+HANDLE_LIBCALL(EXP2_F128, "exp2f128")
 HANDLE_LIBCALL(EXP2_PPCF128, "exp2l")
 HANDLE_LIBCALL(EXP2_FINITE_F32, "__exp2f_finite")
 HANDLE_LIBCALL(EXP2_FINITE_F64, "__exp2_finite")
 HANDLE_LIBCALL(EXP2_FINITE_F80, "__exp2l_finite")
-HANDLE_LIBCALL(EXP2_FINITE_F128, "__exp2l_finite")
+HANDLE_LIBCALL(EXP2_FINITE_F128, "__exp2f128_finite")
 HANDLE_LIBCALL(EXP2_FINITE_PPCF128, "__exp2l_finite")
 HANDLE_LIBCALL(EXP10_F32, "exp10f")
 HANDLE_LIBCALL(EXP10_F64, "exp10")
@@ -190,12 +190,12 @@ HANDLE_LIBCALL(EXP10_PPCF128, "exp10l")
 HANDLE_LIBCALL(SIN_F32, "sinf")
 HANDLE_LIBCALL(SIN_F64, "sin")
 HANDLE_LIBCALL(SIN_F80, "sinl")
-HANDLE_LIBCALL(SIN_F128, "sinl")
+HANDLE_LIBCALL(SIN_F128, "sinf128")
 HANDLE_LIBCALL(SIN_PPCF128, "sinl")
 HANDLE_LIBCALL(COS_F32, "cosf")
 HANDLE_LIBCALL(COS_F64, "cos")
 HANDLE_LIBCALL(COS_F80, "cosl")
-HANDLE_LIBCALL(COS_F128, "cosl")
+HANDLE_LIBCALL(COS_F128, "cosf128")
 HANDLE_LIBCALL(COS_PPCF128, "cosl")
 HANDLE_LIBCALL(SINCOS_F32, nullptr)
 HANDLE_LIBCALL(SINCOS_F64, nullptr)
@@ -207,92 +207,92 @@ HANDLE_LIBCALL(SINCOS_STRET_F64, nullptr)
 HANDLE_LIBCALL(POW_F32, "powf")
 HANDLE_LIBCALL(POW_F64, "pow")
 HANDLE_LIBCALL(POW_F80, "powl")
-HANDLE_LIBCALL(POW_F128, "powl")
+HANDLE_LIBCALL(POW_F128, "powf128")
 HANDLE_LIBCALL(POW_PPCF128, "powl")
 HANDLE_LIBCALL(POW_FINITE_F32, "__powf_finite")
 HANDLE_LIBCALL(POW_FINITE_F64, "__pow_finite")
 HANDLE_LIBCALL(POW_FINITE_F80, "__powl_finite")
-HANDLE_LIBCALL(POW_FINITE_F128, "__powl_finite")
+HANDLE_LIBCALL(POW_FINITE_F128, "__powf128_finite")
 HANDLE_LIBCALL(POW_FINITE_PPCF128, "__powl_finite")
 HANDLE_LIBCALL(CEIL_F32, "ceilf")
 HANDLE_LIBCALL(CEIL_F64, "ceil")
 HANDLE_LIBCALL(CEIL_F80, "ceill")
-HANDLE_LIBCALL(CEIL_F128, "ceill")
+HANDLE_LIBCALL(CEIL_F128, "ceilf128")
 HANDLE_LIBCALL(CEIL_PPCF128, "ceill")
 HANDLE_LIBCALL(TRUNC_F32, "truncf")
 HANDLE_LIBCALL(TRUNC_F64, "trunc")
 HANDLE_LIBCALL(TRUNC_F80, "truncl")
-HANDLE_LIBCALL(TRUNC_F128, "truncl")
+HANDLE_LIBCALL(TRUNC_F128, "truncf128")
 HANDLE_LIBCALL(TRUNC_PPCF128, "truncl")
 HANDLE_LIBCALL(RINT_F32, "rintf")
 HANDLE_LIBCALL(RINT_F64, "rint")
 HANDLE_LIBCALL(RINT_F80, "rintl")
-HANDLE_LIBCALL(RINT_F128, "rintl")
+HANDLE_LIBCALL(RINT_F128, "rintf128")
 HANDLE_LIBCALL(RINT_PPCF128, "rintl")
 HANDLE_LIBCALL(NEARBYINT_F32, "nearbyintf")
 HANDLE_LIBCALL(NEARBYINT_F64, "nearbyint")
 HANDLE_LIBCALL(NEARBYINT_F80, "nearbyintl")
-HANDLE_LIBCALL(NEARBYINT_F128, "nearbyintl")
+HANDLE_LIBCALL(NEARBYINT_F128, "nearbyintf128")
 HANDLE_LIBCALL(NEARBYINT_PPCF128, "nearbyintl")
 HANDLE_LIBCALL(ROUND_F32, "roundf")
 HANDLE_LIBCALL(ROUND_F64, "round")
 HANDLE_LIBCALL(ROUND_F80, "roundl")
-HANDLE_LIBCALL(ROUND_F128, "roundl")
+HANDLE_LIBCALL(ROUND_F128, "roundf128")
 HANDLE_LIBCALL(ROUND_PPCF128, "roundl")
 HANDLE_LIBCALL(ROUNDEVEN_F32, "roundevenf")
 HANDLE_LIBCALL(ROUNDEVEN_F64, "roundeven")
 HANDLE_LIBCALL(ROUNDEVEN_F80, "roundevenl")
-HANDLE_LIBCALL(ROUNDEVEN_F128, "roundevenl")
+HANDLE_LIBCALL(ROUNDEVEN_F128, "roundevenf128")
 HANDLE_LIBCALL(ROUNDEVEN_PPCF128, "roundevenl")
 HANDLE_LIBCALL(FLOOR_F32, "floorf")
 HANDLE_LIBCALL(FLOOR_F64, "floor")
 HANDLE_LIBCALL(FLOOR_F80, "floorl")
-HANDLE_LIBCALL(FLOOR_F128, "floorl")
+HANDLE_LIBCALL(FLOOR_F128, "floorf128")
 HANDLE_LIBCALL(FLOOR_PPCF128, "floorl")
 HANDLE_LIBCALL(COPYSIGN_F32, "copysignf")
 HANDLE_LIBCALL(COPYSIGN_F64, "copysign")
 HANDLE_LIBCALL(COPYSIGN_F80, "copysignl")
-HANDLE_LIBCALL(COPYSIGN_F128, "copysignl")
+HANDLE_LIBCALL(COPYSIGN_F128, "copysignf128")
 HANDLE_LIBCALL(COPYSIGN_PPCF128, "copysignl")
 HANDLE_LIBCALL(FMIN_F32, "fminf")
 HANDLE_LIBCALL(FMIN_F64, "fmin")
 HANDLE_LIBCALL(FMIN_F80, "fminl")
-HANDLE_LIBCALL(FMIN_F128, "fminl")
+HANDLE_LIBCALL(FMIN_F128, "fminf128")
 HANDLE_LIBCALL(FMIN_PPCF128, "fminl")
 HANDLE_LIBCALL(FMAX_F32, "fmaxf")
 HANDLE_LIBCALL(FMAX_F64, "fmax")
 HANDLE_LIBCALL(FMAX_F80, "fmaxl")
-HANDLE_LIBCALL(FMAX_F128, "fmaxl")
+HANDLE_LIBCALL(FMAX_F128, "fmaxf128")
 HANDLE_LIBCALL(FMAX_PPCF128, "fmaxl")
 HANDLE_LIBCALL(LROUND_F32, "lroundf")
 HANDLE_LIBCALL(LROUND_F64, "lround")
 HANDLE_LIBCALL(LROUND_F80, "lroundl")
-HANDLE_LIBCALL(LROUND_F128, "lroundl")
+HANDLE_LIBCALL(LROUND_F128, "lroundf128")
 HANDLE_LIBCALL(LROUND_PPCF128, "lroundl")
 HANDLE_LIBCALL(LLROUND_F32, "llroundf")
 HANDLE_LIBCALL(LLROUND_F64, "llround")
 HANDLE_LIBCALL(LLROUND_F80, "llroundl")
-HANDLE_LIBCALL(LLROUND_F128, "llroundl")
+HANDLE_LIBCALL(LLROUND_F128, "llroundf128")
 HANDLE_LIBCALL(LLROUND_PPCF128, "llroundl")
 HANDLE_LIBCALL(LRINT_F32, "lrintf")
 HANDLE_LIBCALL(LRINT_F64, "lrint")
 HANDLE_LIBCALL(LRINT_F80, "lrintl")
-HANDLE_LIBCALL(LRINT_F128, "lrintl")
+HANDLE_LIBCALL(LRINT_F128, "lrintf128")
 HANDLE_LIBCALL(LRINT_PPCF128, "lrintl")
 HANDLE_LIBCALL(LLRINT_F32, "llrintf")
 HANDLE_LIBCALL(LLRINT_F64, "llrint")
 HANDLE_LIBCALL(LLRINT_F80, "llrintl")
-HANDLE_LIBCALL(LLRINT_F128, "llrintl")
+HANDLE_LIBCALL(LLRINT_F128, "llrintf128")
 HANDLE_LIBCALL(LLRINT_PPCF128, "llrintl")
 HANDLE_LIBCALL(LDEXP_F32, "ldexpf")
 HANDLE_LIBCALL(LDEXP_F64, "ldexp")
 HANDLE_LIBCALL(LDEXP_F80, "ldexpl")
-HANDLE_LIBCALL(LDEXP_F128, "ldexpl")
+HANDLE_LIBCALL(LDEXP_F128, "ldexpf128")
 HANDLE_LIBCALL(LDEXP_PPCF128, "ldexpl")
 HANDLE_LIBCALL(FREXP_F32, "frexpf")
 HANDLE_LIBCALL(FREXP_F64, "frexp")
 HANDLE_LIBCALL(FREXP_F80, "frexpl")
-HANDLE_LIBCALL(FREXP_F128, "frexpl")
+HANDLE_LIBCALL(FREXP_F128, "frexpf128")
 HANDLE_LIBCALL(FREXP_PPCF128, "frexpl")
 
 // Floating point environment
diff --git a/llvm/test/CodeGen/Generic/f128-math-lowering.ll b/llvm/test/CodeGen/Generic/f128-math-lowering.ll
index 8a70786d97fe67..530abb34cbdee9 100644
--- a/llvm/test/CodeGen/Generic/f128-math-lowering.ll
+++ b/llvm/test/CodeGen/Generic/f128-math-lowering.ll
@@ -33,11 +33,11 @@ declare fp128 @llvm.cbrt.f128(fp128)
 
 define fp128 @test_ceilf128(fp128 %a) {
 ; CHECK-LABEL:      test_ceilf128:
-; CHECK-AARCH64:    b ceill
-; CHECK-RISCV32:    call ceill at plt
-; CHECK-S390X:      brasl {{%.*}} ceill at PLT
-; CHECK-X64:        jmp ceill at PLT
-; CHECK-X86:        calll ceill
+; CHECK-AARCH64:    b ceilf128
+; CHECK-RISCV32:    call ceilf128 at plt
+; CHECK-S390X:      brasl {{%.*}} ceilf128 at PLT
+; CHECK-X64:        jmp ceilf128 at PLT
+; CHECK-X86:        calll ceilf128
 start:
   %0 = tail call fp128 @llvm.ceil.f128(fp128 %a)
   ret fp128 %0
@@ -138,11 +138,11 @@ declare fp128 @llvm.copysign.f128(fp128, fp128)
 
 define fp128 @test_cosf128(fp128 %a) {
 ; CHECK-LABEL:      test_cosf128:
-; CHECK-AARCH64:    b cosl
-; CHECK-RISCV32:    call cosl at plt
-; CHECK-S390X:      brasl {{%.*}} cosl at PLT
-; CHECK-X64:        jmp cosl at PLT
-; CHECK-X86:        calll cosl
+; CHECK-AARCH64:    b cosf128
+; CHECK-RISCV32:    call cosf128 at plt
+; CHECK-S390X:      brasl {{%.*}} cosf128 at PLT
+; CHECK-X64:        jmp cosf128 at PLT
+; CHECK-X86:        calll cosf128
 start:
   %0 = tail call fp128 @llvm.cos.f128(fp128 %a)
   ret fp128 %0
@@ -153,11 +153,11 @@ declare fp128 @llvm.cos.f128(fp128)
 
 define fp128 @test_exp2f128(fp128 %a) {
 ; CHECK-LABEL:      test_exp2f128:
-; CHECK-AARCH64:    b exp2l
-; CHECK-RISCV32:    call exp2l at plt
-; CHECK-S390X:      brasl {{%.*}} exp2l at PLT
-; CHECK-X64:        jmp exp2l at PLT
-; CHECK-X86:        calll exp2l
+; CHECK-AARCH64:    b exp2f128
+; CHECK-RISCV32:    call exp2f128 at plt
+; CHECK-S390X:      brasl {{%.*}} exp2f128 at PLT
+; CHECK-X64:        jmp exp2f128 at PLT
+; CHECK-X86:        calll exp2f128
 start:
   %0 = tail call fp128 @llvm.exp2.f128(fp128 %a)
   ret fp128 %0
@@ -183,11 +183,11 @@ declare fp128 @llvm.__exp2f128_finite.f128(fp128)
 
 define fp128 @test_expf128(fp128 %a) {
 ; CHECK-LABEL:      test_expf128:
-; CHECK-AARCH64:    b expl
-; CHECK-RISCV32:    call expl at plt
-; CHECK-S390X:      brasl {{%.*}} expl at PLT
-; CHECK-X64:        jmp expl at PLT
-; CHECK-X86:        calll expl
+; CHECK-AARCH64:    b expf128
+; CHECK-RISCV32:    call expf128 at plt
+; CHECK-S390X:      brasl {{%.*}} expf128 at PLT
+; CHECK-X64:        jmp expf128 at PLT
+; CHECK-X86:        calll expf128
 start:
   %0 = tail call fp128 @llvm.exp.f128(fp128 %a)
   ret fp128 %0
@@ -213,11 +213,11 @@ declare fp128 @llvm.__expf128_finite.f128(fp128)
 
 define fp128 @test_floorf128(fp128 %a) {
 ; CHECK-LABEL:      test_floorf128:
-; CHECK-AARCH64:    b floorl
-; CHECK-RISCV32:    call floorl at plt
-; CHECK-S390X:      brasl {{%.*}} floorl at PLT
-; CHECK-X64:        jmp floorl at PLT
-; CHECK-X86:        calll floorl
+; CHECK-AARCH64:    b floorf128
+; CHECK-RISCV32:    call floorf128 at plt
+; CHECK-S390X:      brasl {{%.*}} floorf128 at PLT
+; CHECK-X64:        jmp floorf128 at PLT
+; CHECK-X86:        calll floorf128
 start:
   %0 = tail call fp128 @llvm.floor.f128(fp128 %a)
   ret fp128 %0
@@ -228,11 +228,11 @@ declare fp128 @llvm.floor.f128(fp128)
 
 define fp128 @test_fmaf128(fp128 %a, fp128 %b, fp128 %c) {
 ; CHECK-LABEL:      test_fmaf128:
-; CHECK-AARCH64:    b fmal
-; CHECK-RISCV32:    call fmal at plt
-; CHECK-S390X:      brasl {{%.*}} fmal at PLT
-; CHECK-X64:        jmp fmal at PLT
-; CHECK-X86:        calll fmal
+; CHECK-AARCH64:    b fmaf128
+; CHECK-RISCV32:    call fmaf128 at plt
+; CHECK-S390X:      brasl {{%.*}} fmaf128 at PLT
+; CHECK-X64:        jmp fmaf128 at PLT
+; CHECK-X86:        calll fmaf128
 start:
   %0 = tail call fp128 @llvm.fma.f128(fp128 %a, fp128 %b, fp128 %c)
   ret fp128 %0
@@ -288,11 +288,11 @@ declare fp128 @llvm.fmod.f128(fp128, fp128)
 
 define { fp128, i32 } @test_frexpf128(fp128 %a) {
 ; CHECK-LABEL:      test_frexpf128:
-; CHECK-AARCH64:    bl frexpl
-; CHECK-RISCV32:    call frexpl at plt
-; CHECK-S390X:      brasl {{%.*}} frexpl at PLT
-; CHECK-X64:        callq frexpl at PLT
-; CHECK-X86:        calll frexpl
+; CHECK-AARCH64:    bl frexpf128
+; CHECK-RISCV32:    call frexpf128 at plt
+; CHECK-S390X:      brasl {{%.*}} frexpf128 at PLT
+; CHECK-X64:        callq frexpf128 at PLT
+; CHECK-X86:        calll frexpf128
 start:
   %0 = tail call { fp128, i32 } @llvm.frexp.f128(fp128 %a)
   ret { fp128, i32 } %0
@@ -303,11 +303,11 @@ declare { fp128, i32 } @llvm.frexp.f128(fp128)
 
 define fp128 @test_ldexpf128(fp128 %a, i32 %b) {
 ; CHECK-LABEL:      test_ldexpf128:
-; CHECK-AARCH64:    b ldexpl
-; CHECK-RISCV32:    call ldexpl at plt
-; CHECK-S390X:      brasl {{%.*}} ldexpl at PLT
-; CHECK-X64:        jmp ldexpl at PLT
-; CHECK-X86:        calll ldexpl
+; CHECK-AARCH64:    b ldexpf128
+; CHECK-RISCV32:    call ldexpf128 at plt
+; CHECK-S390X:      brasl {{%.*}} ldexpf128 at PLT
+; CHECK-X64:        jmp ldexpf128 at PLT
+; CHECK-X86:        calll ldexpf128
 start:
   %0 = tail call fp128 @llvm.ldexp.f128(fp128 %a, i32 %b)
   ret fp128 %0
@@ -318,11 +318,11 @@ declare fp128 @llvm.ldexp.f128(fp128, i32)
 
 define i64 @test_llrintf128(fp128 %a) {
 ; CHECK-LABEL:      test_llrintf128:
-; CHECK-AARCH64:    b llrintl
-; CHECK-RISCV32:    call llrintl at plt
-; CHECK-S390X:      brasl {{%.*}} llrintl at PLT
-; CHECK-X64:        jmp llrintl at PLT
-; CHECK-X86:        calll llrintl
+; CHECK-AARCH64:    b llrintf128
+; CHECK-RISCV32:    call llrintf128 at plt
+; CHECK-S390X:      brasl {{%.*}} llrintf128 at PLT
+; CHECK-X64:        jmp llrintf128 at PLT
+; CHECK-X86:        calll llrintf128
 start:
   %0 = tail call i64 @llvm.llrint.f128(fp128 %a)
   ret i64 %0
@@ -333,11 +333,11 @@ declare i64 @llvm.llrint.f128(fp128)
 
 define i64 @test_llroundf128(fp128 %a) {
 ; CHECK-LABEL:      test_llroundf128:
-; CHECK-AARCH64:    b llroundl
-; CHECK-RISCV32:    call llroundl at plt
-; CHECK-S390X:      brasl {{%.*}} llroundl at PLT
-; CHECK-X64:        jmp llroundl at PLT
-; CHECK-X86:        calll llroundl
+; CHECK-AARCH64:    b llroundf128
+; CHECK-RISCV32:    call llroundf128 at plt
+; CHECK-S390X:      brasl {{%.*}} llroundf128 at PLT
+; CHECK-X64:        jmp llroundf128 at PLT
+; CHECK-X86:        calll llroundf128
 start:
   %0 = tail call i64 @llvm.llround.i64.f128(fp128 %a)
   ret i64 %0
@@ -348,11 +348,11 @@ declare i64 @llvm.llround.i64.f128(fp128)
 
 define fp128 @test_log10f128(fp128 %a) {
 ; CHECK-LABEL:      test_log10f128:
-; CHECK-AARCH64:    b log10l
-; CHECK-RISCV32:    call log10l at plt
-; CHECK-S390X:      brasl {{%.*}} log10l at PLT
-; CHECK-X64:        jmp log10l at PLT
-; CHECK-X86:        calll log10l
+; CHECK-AARCH64:    b log10f128
+; CHECK-RISCV32:    call log10f128 at plt
+; CHECK-S390X:      brasl {{%.*}} log10f128 at PLT
+; CHECK-X64:        jmp log10f128 at PLT
+; CHECK-X86:        calll log10f128
 start:
   %0 = tail call fp128 @llvm.log10.f128(fp128 %a)
   ret fp128 %0
@@ -378,11 +378,11 @@ declare fp128 @llvm.__log10f128_finite.f128(fp128)
 
 define fp128 @test_log2f128(fp128 %a) {
 ; CHECK-LABEL:      test_log2f128:
-; CHECK-AARCH64:    b log2l
-; CHECK-RISCV32:    call log2l at plt
-; CHECK-S390X:      brasl {{%.*}} log2l at PLT
-; CHECK-X64:        jmp log2l at PLT
-; CHECK-X86:        calll log2l
+; CHECK-AARCH64:    b log2f128
+; CHECK-RISCV32:    call log2f128 at plt
+; CHECK-S390X:      brasl {{%.*}} log2f128 at PLT
+; CHECK-X64:        jmp log2f128 at PLT
+; CHECK-X86:        calll log2f128
 start:
   %0 = tail call fp128 @llvm.log2.f128(fp128 %a)
   ret fp128 %0
@@ -408,11 +408,11 @@ declare fp128 @llvm.__log2f128_finite.f128(fp128)
 
 define fp128 @test_logf128(fp128 %a) {
 ; CHECK-LABEL:      test_logf128:
-; CHECK-AARCH64:    b logl
-; CHECK-RISCV32:    call logl at plt
-; CHECK-S390X:      brasl {{%.*}} logl at PLT
-; CHECK-X64:        jmp logl at PLT
-; CHECK-X86:        calll logl
+; CHECK-AARCH64:    b logf128
+; CHECK-RISCV32:    call logf128 at plt
+; CHECK-S390X:      brasl {{%.*}} logf128 at PLT
+; CHECK-X64:        jmp logf128 at PLT
+; CHECK-X86:        calll logf128
 start:
   %0 = tail call fp128 @llvm.log.f128(fp128 %a)
   ret fp128 %0
@@ -438,11 +438,11 @@ declare fp128 @llvm.__logf128_finite.f128(fp128)
 
 define i64 @test_lrintf128(fp128 %a) {
 ; CHECK-LABEL:      test_lrintf128:
-; CHECK-AARCH64:    b lrintl
-; CHECK-RISCV32:    call lrintl at plt
-; CHECK-S390X:      brasl {{%.*}} lrintl at PLT
-; CHECK-X64:        jmp lrintl at PLT
-; CHECK-X86:        calll lrintl
+; CHECK-AARCH64:    b lrintf128
+; CHECK-RISCV32:    call lrintf128 at plt
+; CHECK-S390X:      brasl {{%.*}} lrintf128 at PLT
+; CHECK-X64:        jmp lrintf128 at PLT
+; CHECK-X86:        calll lrintf128
 start:
   %0 = tail call i64 @llvm.lrint.f128(fp128 %a)
   ret i64 %0
@@ -453,11 +453,11 @@ declare i64 @llvm.lrint.f128(fp128)
 
 define i64 @test_lroundf128(fp128 %a) {
 ; CHECK-LABEL:      test_lroundf128:
-; CHECK-AARCH64:    b lroundl
-; CHECK-RISCV32:    call lroundl at plt
-; CHECK-S390X:      brasl {{%.*}} lroundl at PLT
-; CHECK-X64:        jmp lroundl at PLT
-; CHECK-X86:        calll lroundl
+; CHECK-AARCH64:    b lroundf128
+; CHECK-RISCV32:    call lroundf128 at plt
+; CHECK-S390X:      brasl {{%.*}} lroundf128 at PLT
+; CHECK-X64:        jmp lroundf128 at PLT
+; CHECK-X86:        calll lroundf128
 start:
   %0 = tail call i64 @llvm.lround.i64.f128(fp128 %a)
   ret i64 %0
@@ -468,11 +468,11 @@ declare i64 @llvm.lround.i64.f128(fp128)
 
 define fp128 @test_nearbyintf128(fp128 %a) {
 ; CHECK-LABEL:      test_nearbyintf128:
-; CHECK-AARCH64:    b nearbyintl
-; CHECK-RISCV32:    call nearbyintl at plt
-; CHECK-S390X:      brasl {{%.*}} nearbyintl at PLT
-; CHECK-X64:        jmp nearbyintl at PLT
-; CHECK-X86:        calll nearbyintl
+; CHECK-AARCH64:    b nearbyintf128
+; CHECK-RISCV32:    call nearbyintf128 at plt
+; CHECK-S390X:      brasl {{%.*}} nearbyintf128 at PLT
+; CHECK-X64:        jmp nearbyintf128 at PLT
+; CHECK-X86:        calll nearbyintf128
 start:
   %0 = tail call fp128 @llvm.nearbyint.f128(fp128 %a)
   ret fp128 %0
@@ -483,11 +483,11 @@ declare fp128 @llvm.nearbyint.f128(fp128)
 
 define fp128 @test_powf128(fp128 %a, fp128 %b) {
 ; CHECK-LABEL:      test_powf128:
-; CHECK-AARCH64:    b powl
-; CHECK-RISCV32:    call powl at plt
-; CHECK-S390X:      brasl {{%.*}} powl at PLT
-; CHECK-X64:        jmp powl at PLT
-; CHECK-X86:        calll powl
+; CHECK-AARCH64:    b powf128
+; CHECK-RISCV32:    call powf128 at plt
+; CHECK-S390X:      brasl {{%.*}} powf128 at PLT
+; CHECK-X64:        jmp powf128 at PLT
+; CHECK-X86:        calll powf128
 start:
   %0 = tail call fp128 @llvm.pow.f128(fp128 %a, fp128 %b)
   ret fp128 %0
@@ -513,8 +513,8 @@ declare fp128 @llvm.__powf128_finite.f128(fp128, fp128)
 
 define fp128 @test_rintf128(fp128 %a) {
 ; CHECK-LABEL:      test_rintf128:
-; CHECK-AARCH64:    b rintl
-; CHECK-RISCV32:    call rintl at plt
+; CHECK-AARCH64:    b rintf128
+; CHECK-RISCV32:    call rintf128 at plt
 ;
 ; CHECK-S390X-LABEL: test_rintf128:
 ; CHECK-S390X:       # %bb.0: # %start
@@ -525,8 +525,8 @@ define fp128 @test_rintf128(fp128 %a) {
 ; CHECK-S390X-NEXT:    std %f2, 8(%r2)
 ; CHECK-S390X-NEXT:    br %r14
 ;
-; CHECK-X64:        jmp rintl at PLT
-; CHECK-X86:        calll rintl
+; CHECK-X64:        jmp rintf128 at PLT
+; CHECK-X86:        calll rintf128
 start:
   %0 = tail call fp128 @llvm.rint.f128(fp128 %a)
   ret fp128 %0
@@ -537,11 +537,11 @@ declare fp128 @llvm.rint.f128(fp128)
 
 define fp128 @test_roundevenf128(fp128 %a) {
 ; CHECK-LABEL:      test_roundevenf128:
-; CHECK-AARCH64:    b roundevenl
-; CHECK-RISCV32:    call roundevenl at plt
-; CHECK-S390X:      brasl {{%.*}} roundevenl at PLT
-; CHECK-X64:        jmp roundevenl at PLT
-; CHECK-X86:        calll roundevenl
+; CHECK-AARCH64:    b roundevenf128
+; CHECK-RISCV32:    call roundevenf128 at plt
+; CHECK-S390X:      brasl {{%.*}} roundevenf128 at PLT
+; CHECK-X64:        jmp roundevenf128 at PLT
+; CHECK-X86:        calll roundevenf128
 start:
   %0 = tail call fp128 @llvm.roundeven.f128(fp128 %a)
   ret fp128 %0
@@ -552,11 +552,11 @@ declare fp128 @llvm.roundeven.f128(fp128)
 
 define fp128 @test_roundf128(fp128 %a) {
 ; CHECK-LABEL:      test_roundf128:
-; CHECK-AARCH64:    b roundl
-; CHECK-RISCV32:    call roundl at plt
-; CHECK-S390X:      brasl {{%.*}} roundl at PLT
-; CHECK-X64:        jmp roundl at PLT
-; CHECK-X86:        calll roundl
+; CHECK-AARCH64:    b roundf128
+; CHECK-RISCV32:    call roundf128 at plt
+; CHECK-S390X:      brasl {{%.*}} roundf128 at PLT
+; CHECK-X64:        jmp roundf128 at PLT
+; CHECK-X86:        calll roundf128
 start:
   %0 = tail call fp128 @llvm.round.f128(fp128 %a)
   ret fp128 %0
@@ -567,11 +567,11 @@ declare fp128 @llvm.round.f128(fp128)
 
 define fp128 @test_sinf128(fp128 %a) {
 ; CHECK-LABEL:      test_sinf128:
-; CHECK-AARCH64:    b sinl
-; CHECK-RISCV32:    call sinl at plt
-; CHECK-S390X:      brasl {{%.*}} sinl at PLT
-; CHECK-X64:        jmp sinl at PLT
-; CHECK-X86:        calll sinl
+; CHECK-AARCH64:    b sinf128
+; CHECK-RISCV32:    call sinf128 at plt
+; CHECK-S390X:      brasl {{%.*}} sinf128 at PLT
+; CHECK-X64:        jmp sinf128 at PLT
+; CHECK-X86:        calll sinf128
 start:
   %0 = tail call fp128 @llvm.sin.f128(fp128 %a)
   ret fp128 %0
@@ -582,11 +582,11 @@ declare fp128 @llvm.sin.f128(fp128)
 
 define fp128 @test_sqrtf128(fp128 %a) {
 ; CHECK-LABEL:      test_sqrtf128:
-; CHECK-AARCH64:    b sqrtl
-; CHECK-RISCV32:    call sqrtl at plt
+; CHECK-AARCH64:    b sqrtf128
+; CHECK-RISCV32:    call sqrtf128 at plt
 ; CHECK-S390X:      sqxbr {{%.*}} {{%.*}}
-; CHECK-X64:        jmp sqrtl at PLT
-; CHECK-X86:        calll sqrtl
+; CHECK-X64:        jmp sqrtf128 at PLT
+; CHECK-X86:        calll sqrtf128
 start:
   %0 = tail call fp128 @llvm.sqrt.f128(fp128 %a)
   ret fp128 %0
@@ -597,11 +597,11 @@ declare fp128 @llvm.sqrt.f128(fp128)
 
 define fp128 @test_truncf128(fp128 %a) {
 ; CHECK-LABEL:      test_truncf128:
-; CHECK-AARCH64:    b truncl
-; CHECK-RISCV32:    call truncl at plt
-; CHECK-S390X:      brasl {{%.*}} truncl at PLT
-; CHECK-X64:        jmp truncl at PLT
-; CHECK-X86:        calll truncl
+; CHECK-AARCH64:    b truncf128
+; CHECK-RISCV32:    call truncf128 at plt
+; CHECK-S390X:      brasl {{%.*}} truncf128 at PLT
+; CHECK-X64:        jmp truncf128 at PLT
+; CHECK-X86:        calll truncf128
 start:
   %0 = tail call fp128 @llvm.trunc.f128(fp128 %a)
   ret fp128 %0
diff --git a/llvm/test/CodeGen/X86/f128-arith.ll b/llvm/test/CodeGen/X86/f128-arith.ll
new file mode 100644
index 00000000000000..a65d9039cb69b2
--- /dev/null
+++ b/llvm/test/CodeGen/X86/f128-arith.ll
@@ -0,0 +1,515 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 2
+; RUN: llc < %s -mtriple=i686-unknown-unknown   -verify-machineinstrs | FileCheck %s --check-prefix=CHECK-32
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -verify-machineinstrs | FileCheck %s --check-prefix=CHECK-64
+;
+; Test lowering of fp128 intrinsics
+
+define fp128 @test_cbrtf128(fp128 %a) {
+; CHECK-32-LABEL: test_cbrtf128:
+; CHECK-32:    calll llvm.cbrt.f128 at PLT
+;
+; CHECK-64-LABEL: test_cbrtf128:
+; CHECK-64:    jmp llvm.cbrt.f128 at PLT # TAILCALL
+start:
+  %0 = tail call fp128 @llvm.cbrt.f128(fp128 %a)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.cbrt.f128(fp128)
+
+
+define fp128 @test_ceilf128(fp128 %a) {
+; CHECK-32-LABEL: test_ceilf128:
+; CHECK-32:    calll ceilf128
+;
+; CHECK-64-LABEL: test_ceilf128:
+; CHECK-64:    jmp ceilf128 at PLT # TAILCALL
+start:
+  %0 = tail call fp128 @llvm.ceil.f128(fp128 %a)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.ceil.f128(fp128)
+
+
+define fp128 @test_copysignf128(fp128 %a, fp128 %b) {
+; CHECK-32-LABEL: test_copysignf128:
+; CHECK-32:       # %bb.0: # %start
+; CHECK-32-NEXT:    pushl %ebx
+; CHECK-32-NEXT:    .cfi_def_cfa_offset 8
+; CHECK-32-NEXT:    pushl %edi
+; CHECK-32-NEXT:    .cfi_def_cfa_offset 12
+; CHECK-32-NEXT:    pushl %esi
+; CHECK-32-NEXT:    .cfi_def_cfa_offset 16
+; CHECK-32-NEXT:    .cfi_offset %esi, -16
+; CHECK-32-NEXT:    .cfi_offset %edi, -12
+; CHECK-32-NEXT:    .cfi_offset %ebx, -8
+; CHECK-32-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; CHECK-32-NEXT:    movl {{[0-9]+}}(%esp), %ecx
+; CHECK-32-NEXT:    movl {{[0-9]+}}(%esp), %edx
+; CHECK-32-NEXT:    movl {{[0-9]+}}(%esp), %esi
+; CHECK-32-NEXT:    movl $-2147483648, %edi # imm = 0x80000000
+; CHECK-32-NEXT:    andl {{[0-9]+}}(%esp), %edi
+; CHECK-32-NEXT:    movl $2147483647, %ebx # imm = 0x7FFFFFFF
+; CHECK-32-NEXT:    andl {{[0-9]+}}(%esp), %ebx
+; CHECK-32-NEXT:    orl %edi, %ebx
+; CHECK-32-NEXT:    movl %ebx, 12(%eax)
+; CHECK-32-NEXT:    movl %esi, 8(%eax)
+; CHECK-32-NEXT:    movl %edx, 4(%eax)
+; CHECK-32-NEXT:    movl %ecx, (%eax)
+; CHECK-32-NEXT:    popl %esi
+; CHECK-32-NEXT:    .cfi_def_cfa_offset 12
+; CHECK-32-NEXT:    popl %edi
+; CHECK-32-NEXT:    .cfi_def_cfa_offset 8
+; CHECK-32-NEXT:    popl %ebx
+; CHECK-32-NEXT:    .cfi_def_cfa_offset 4
+; CHECK-32-NEXT:    retl $4
+;
+; CHECK-64-LABEL: test_copysignf128:
+; CHECK-64:       # %bb.0: # %start
+; CHECK-64-NEXT:    andps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1
+; CHECK-64-NEXT:    andps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
+; CHECK-64-NEXT:    orps %xmm1, %xmm0
+; CHECK-64-NEXT:    retq
+; FIXME: calling long double rather than f128 function
+; FIXME: calling long double rather than f128 function
+start:
+  %0 = tail call fp128 @llvm.copysign.f128(fp128 %a, fp128 %b)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.copysign.f128(fp128, fp128)
+
+
+define fp128 @test_cosf128(fp128 %a) {
+; CHECK-32-LABEL: test_cosf128:
+; CHECK-32:    calll cosf128
+;
+; CHECK-64-LABEL: test_cosf128:
+; CHECK-64:    jmp cosf128 at PLT # TAILCALL
+start:
+  %0 = tail call fp128 @llvm.cos.f128(fp128 %a)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.cos.f128(fp128)
+
+
+define fp128 @test_exp2f128(fp128 %a) {
+; CHECK-32-LABEL: test_exp2f128:
+; CHECK-32:    calll exp2f128
+;
+; CHECK-64-LABEL: test_exp2f128:
+; CHECK-64:    jmp exp2f128 at PLT # TAILCALL
+start:
+  %0 = tail call fp128 @llvm.exp2.f128(fp128 %a)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.exp2.f128(fp128)
+
+
+define fp128 @test___exp2f128_finite(fp128 %a) {
+; CHECK-32-LABEL: test___exp2f128_finite:
+; CHECK-32:    calll llvm.__exp2f128_finite.f128 at PLT
+;
+; CHECK-64-LABEL: test___exp2f128_finite:
+; CHECK-64:    jmp llvm.__exp2f128_finite.f128 at PLT # TAILCALL
+start:
+  %0 = tail call fp128 @llvm.__exp2f128_finite.f128(fp128  %a)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.__exp2f128_finite.f128(fp128)
+
+
+define fp128 @test_expf128(fp128 %a) {
+; CHECK-32-LABEL: test_expf128:
+; CHECK-32:    calll expf128
+;
+; CHECK-64-LABEL: test_expf128:
+; CHECK-64:    jmp expf128 at PLT # TAILCALL
+start:
+  %0 = tail call fp128 @llvm.exp.f128(fp128 %a)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.exp.f128(fp128)
+
+
+define fp128 @test___expf128_finite(fp128 %a) {
+; CHECK-32-LABEL: test___expf128_finite:
+; CHECK-32:    calll llvm.__expf128_finite.f128 at PLT
+;
+; CHECK-64-LABEL: test___expf128_finite:
+; CHECK-64:    jmp llvm.__expf128_finite.f128 at PLT # TAILCALL
+start:
+  %0 = tail call fp128 @llvm.__expf128_finite.f128(fp128 %a)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.__expf128_finite.f128(fp128)
+
+
+define fp128 @test_floorf128(fp128 %a) {
+; CHECK-32-LABEL: test_floorf128:
+; CHECK-32:    calll floorf128
+;
+; CHECK-64-LABEL: test_floorf128:
+; CHECK-64:    jmp floorf128 at PLT # TAILCALL
+start:
+  %0 = tail call fp128 @llvm.floor.f128(fp128 %a)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.floor.f128(fp128)
+
+
+define fp128 @test_fmaf128(fp128 %a, fp128 %b, fp128 %c) {
+; CHECK-32-LABEL: test_fmaf128:
+; CHECK-32:    calll fmaf128
+;
+; CHECK-64-LABEL: test_fmaf128:
+; CHECK-64:    jmp fmaf128 at PLT # TAILCALL
+start:
+  %0 = tail call fp128 @llvm.fma.f128(fp128 %a, fp128 %b, fp128 %c)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.fma.f128(fp128, fp128, fp128)
+
+
+define fp128 @test_fmaxf128(fp128 %a, fp128 %b) {
+; CHECK-32-LABEL: test_fmaxf128:
+; CHECK-32:    calll llvm.fmax.f128 at PLT
+;
+; CHECK-64-LABEL: test_fmaxf128:
+; CHECK-64:    jmp llvm.fmax.f128 at PLT # TAILCALL
+start:
+  %0 = tail call fp128 @llvm.fmax.f128(fp128 %a, fp128 %b)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.fmax.f128(fp128, fp128)
+
+
+define fp128 @test_fminf128(fp128 %a, fp128 %b) {
+; CHECK-32-LABEL: test_fminf128:
+; CHECK-32:    calll llvm.fmin.f128 at PLT
+;
+; CHECK-64-LABEL: test_fminf128:
+; CHECK-64:    jmp llvm.fmin.f128 at PLT # TAILCALL
+start:
+  %0 = tail call fp128 @llvm.fmin.f128(fp128 %a, fp128 %b)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.fmin.f128(fp128, fp128)
+
+
+define fp128 @test_fmodf128(fp128 %a, fp128 %b) {
+; CHECK-32-LABEL: test_fmodf128:
+; CHECK-32:    calll llvm.fmod.f128 at PLT
+;
+; CHECK-64-LABEL: test_fmodf128:
+; CHECK-64:    jmp llvm.fmod.f128 at PLT # TAILCALL
+start:
+  %0 = tail call fp128 @llvm.fmod.f128(fp128 %a, fp128 %b)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.fmod.f128(fp128, fp128)
+
+
+define { fp128, i32 } @test_frexpf128(fp128 %a) {
+; CHECK-32-LABEL: test_frexpf128:
+; CHECK-32:    calll frexpf128
+;
+; CHECK-64-LABEL: test_frexpf128:
+; CHECK-64:    callq frexpf128 at PLT
+start:
+  %0 = tail call { fp128, i32 } @llvm.frexp.f128(fp128 %a)
+  ret { fp128, i32 } %0
+}
+
+declare { fp128, i32 } @llvm.frexp.f128(fp128)
+
+
+define fp128 @test_ldexpf128(fp128 %a, i32 %b) {
+; CHECK-32-LABEL: test_ldexpf128:
+; CHECK-32:    calll ldexpf128
+;
+; CHECK-64-LABEL: test_ldexpf128:
+; CHECK-64:    jmp ldexpf128 at PLT # TAILCALL
+start:
+  %0 = tail call fp128 @llvm.ldexp.f128(fp128 %a, i32 %b)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.ldexp.f128(fp128, i32)
+
+
+define i64 @test_llrintf128(fp128 %a) {
+; CHECK-32-LABEL: test_llrintf128:
+; CHECK-32:    calll llrintf128
+;
+; CHECK-64-LABEL: test_llrintf128:
+; CHECK-64:    jmp llrintf128 at PLT # TAILCALL
+start:
+  %0 = tail call i64 @llvm.llrint.f128(fp128 %a)
+  ret i64 %0
+}
+
+declare i64 @llvm.llrint.f128(fp128)
+
+
+define i64 @test_llroundf128(fp128 %a) {
+; CHECK-32-LABEL: test_llroundf128:
+; CHECK-32:    calll llroundf128
+;
+; CHECK-64-LABEL: test_llroundf128:
+; CHECK-64:    jmp llroundf128 at PLT # TAILCALL
+start:
+  %0 = tail call i64 @llvm.llround.i64.f128(fp128 %a)
+  ret i64 %0
+}
+
+declare i64 @llvm.llround.i64.f128(fp128)
+
+
+define fp128 @test_log10f128(fp128 %a) {
+; CHECK-32-LABEL: test_log10f128:
+; CHECK-32:    calll log10f128
+;
+; CHECK-64-LABEL: test_log10f128:
+; CHECK-64:    jmp log10f128 at PLT # TAILCALL
+start:
+  %0 = tail call fp128 @llvm.log10.f128(fp128 %a)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.log10.f128(fp128)
+
+
+define fp128 @test___log10f128_finite(fp128 %a) {
+; CHECK-32-LABEL: test___log10f128_finite:
+; CHECK-32:    calll llvm.__log10f128_finite.f128 at PLT
+;
+; CHECK-64-LABEL: test___log10f128_finite:
+; CHECK-64:    jmp llvm.__log10f128_finite.f128 at PLT # TAILCALL
+start:
+  %0 = tail call fp128 @llvm.__log10f128_finite.f128(fp128 %a)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.__log10f128_finite.f128(fp128)
+
+
+define fp128 @test_log2f128(fp128 %a) {
+; CHECK-32-LABEL: test_log2f128:
+; CHECK-32:    calll log2f128
+;
+; CHECK-64-LABEL: test_log2f128:
+; CHECK-64:   jmp log2f128 at PLT # TAILCALL
+start:
+  %0 = tail call fp128 @llvm.log2.f128(fp128 %a)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.log2.f128(fp128)
+
+
+define fp128 @test___log2f128_finite(fp128 %a) {
+; CHECK-32-LABEL: test___log2f128_finite:
+; CHECK-32:    calll llvm.__log2f128_finite.f128 at PLT
+;
+; CHECK-64-LABEL: test___log2f128_finite:
+; CHECK-64:    jmp llvm.__log2f128_finite.f128 at PLT # TAILCALL
+start:
+  %0 = tail call fp128 @llvm.__log2f128_finite.f128(fp128 %a)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.__log2f128_finite.f128(fp128)
+
+
+define fp128 @test_logf128(fp128 %a) {
+; CHECK-32-LABEL: test_logf128:
+; CHECK-32:    calll logf128
+;
+; CHECK-64-LABEL: test_logf128:
+; CHECK-64:    jmp logf128 at PLT # TAILCALL
+start:
+  %0 = tail call fp128 @llvm.log.f128(fp128 %a)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.log.f128(fp128)
+
+
+define fp128 @test___logf128_finite(fp128 %a) {
+; CHECK-32-LABEL: test___logf128_finite:
+; CHECK-32:    calll llvm.__logf128_finite.f128 at PLT
+;
+; CHECK-64-LABEL: test___logf128_finite:
+; CHECK-64:    jmp llvm.__logf128_finite.f128 at PLT # TAILCALL
+start:
+  %0 = tail call fp128 @llvm.__logf128_finite.f128(fp128 %a)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.__logf128_finite.f128(fp128)
+
+
+define i64 @test_lrintf128(fp128 %a) {
+; CHECK-32-LABEL: test_lrintf128:
+; CHECK-32:    calll lrintf128
+;
+; CHECK-64-LABEL: test_lrintf128:
+; CHECK-64:    jmp lrintf128 at PLT # TAILCALL
+start:
+  %0 = tail call i64 @llvm.lrint.f128(fp128 %a)
+  ret i64 %0
+}
+
+declare i64 @llvm.lrint.f128(fp128)
+
+
+define i64 @test_lroundf128(fp128 %a) {
+; CHECK-32-LABEL: test_lroundf128:
+; CHECK-32:    calll lroundf128
+;
+; CHECK-64-LABEL: test_lroundf128:
+; CHECK-64:    jmp lroundf128 at PLT # TAILCALL
+start:
+  %0 = tail call i64 @llvm.lround.i64.f128(fp128 %a)
+  ret i64 %0
+}
+
+declare i64 @llvm.lround.i64.f128(fp128)
+
+
+define fp128 @test_nearbyintf128(fp128 %a) {
+; CHECK-32-LABEL: test_nearbyintf128:
+; CHECK-32:    calll nearbyintf128
+;
+; CHECK-64-LABEL: test_nearbyintf128:
+; CHECK-64:    jmp nearbyintf128 at PLT # TAILCALL
+start:
+  %0 = tail call fp128 @llvm.nearbyint.f128(fp128 %a)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.nearbyint.f128(fp128)
+
+
+define fp128 @test_powf128(fp128 %a, fp128 %b) {
+; CHECK-32-LABEL: test_powf128:
+; CHECK-32:    calll powf128
+;
+; CHECK-64-LABEL: test_powf128:
+; CHECK-64:    jmp powf128 at PLT # TAILCALL
+start:
+  %0 = tail call fp128 @llvm.pow.f128(fp128 %a, fp128 %b)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.pow.f128(fp128, fp128)
+
+
+define fp128 @test___powf128_finite(fp128 %a, fp128 %b) {
+; CHECK-32-LABEL: test___powf128_finite:
+; CHECK-32:    calll llvm.__powf128_finite.f128 at PLT
+;
+; CHECK-64-LABEL: test___powf128_finite:
+; CHECK-64:    jmp llvm.__powf128_finite.f128 at PLT # TAILCALL
+start:
+  %0 = tail call fp128 @llvm.__powf128_finite.f128(fp128 %a, fp128 %b)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.__powf128_finite.f128(fp128, fp128)
+
+
+define fp128 @test_rintf128(fp128 %a) {
+; CHECK-32-LABEL: test_rintf128:
+; CHECK-32:    calll rintf128
+;
+; CHECK-64-LABEL: test_rintf128:
+; CHECK-64:    jmp rintf128 at PLT # TAILCALL
+start:
+  %0 = tail call fp128 @llvm.rint.f128(fp128 %a)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.rint.f128(fp128)
+
+
+define fp128 @test_roundevenf128(fp128 %a) {
+; CHECK-32-LABEL: test_roundevenf128:
+; CHECK-32:    calll roundevenf128
+;
+; CHECK-64-LABEL: test_roundevenf128:
+; CHECK-64:    jmp roundevenf128 at PLT # TAILCALL
+start:
+  %0 = tail call fp128 @llvm.roundeven.f128(fp128 %a)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.roundeven.f128(fp128)
+
+
+define fp128 @test_roundf128(fp128 %a) {
+; CHECK-32-LABEL: test_roundf128:
+; CHECK-32:    calll roundf128
+;
+; CHECK-64-LABEL: test_roundf128:
+; CHECK-64:    jmp roundf128 at PLT # TAILCALL
+start:
+  %0 = tail call fp128 @llvm.round.f128(fp128 %a)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.round.f128(fp128)
+
+
+define fp128 @test_sinf128(fp128 %a) {
+; CHECK-32-LABEL: test_sinf128:
+; CHECK-32:    calll sinf128
+;
+; CHECK-64-LABEL: test_sinf128:
+; CHECK-64:    jmp sinf128 at PLT # TAILCALL
+start:
+  %0 = tail call fp128 @llvm.sin.f128(fp128 %a)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.sin.f128(fp128)
+
+
+define fp128 @test_sqrtf128(fp128 %a) {
+; CHECK-32-LABEL: test_sqrtf128:
+; CHECK-32:    calll sqrtf128
+;
+; CHECK-64-LABEL: test_sqrtf128:
+; CHECK-64:    jmp sqrtf128 at PLT # TAILCALL
+start:
+  %0 = tail call fp128 @llvm.sqrt.f128(fp128 %a)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.sqrt.f128(fp128)
+
+
+define fp128 @test_truncf128(fp128 %a) {
+; CHECK-32-LABEL: test_truncf128:
+; CHECK-32:    calll truncf128
+;
+; CHECK-64-LABEL: test_truncf128:
+; CHECK-64:    jmp truncf128 at PLT # TAILCALL
+start:
+  %0 = tail call fp128 @llvm.trunc.f128(fp128 %a)
+  ret fp128 %0
+}
+
+declare fp128 @llvm.trunc.f128(fp128)

>From b5d415ec5e5498626d37b7104ef2c4d493b00840 Mon Sep 17 00:00:00 2001
From: Trevor Gross <tmgross at umich.edu>
Date: Fri, 29 Dec 2023 03:45:24 -0500
Subject: [PATCH 4/4] [IR] Use long double functions for `fp128` intrinsics
 where possible

Functions such as `sinf128` are not always available. For targets where `long
double` is `f128`, use functions such as `sinfl` instead.
---
 llvm/lib/CodeGen/TargetLoweringBase.cpp | 38 +++++++++++++++++++++++++
 1 file changed, 38 insertions(+)

diff --git a/llvm/lib/CodeGen/TargetLoweringBase.cpp b/llvm/lib/CodeGen/TargetLoweringBase.cpp
index 2648c16bcd8d90..d59affcde907bb 100644
--- a/llvm/lib/CodeGen/TargetLoweringBase.cpp
+++ b/llvm/lib/CodeGen/TargetLoweringBase.cpp
@@ -221,6 +221,44 @@ void TargetLoweringBase::InitLibcalls(const Triple &TT) {
     setLibcallName(RTLIB::FREXP_F128, nullptr);
     setLibcallName(RTLIB::FREXP_PPCF128, nullptr);
   }
+
+  if (TT.isLongDoubleF128()) {
+    // Use the more available long double functions for fp128 if possible
+    setLibcallName(RTLIB::REM_F128, "fmodl");
+    setLibcallName(RTLIB::FMA_F128, "fmal");
+    setLibcallName(RTLIB::SQRT_F128, "sqrtl");
+    setLibcallName(RTLIB::CBRT_F128, "cbrtl");
+    setLibcallName(RTLIB::LOG_F128, "logl");
+    setLibcallName(RTLIB::LOG_FINITE_F128, "__logl_finite");
+    setLibcallName(RTLIB::LOG2_F128, "log2l");
+    setLibcallName(RTLIB::LOG2_FINITE_F128, "__log2l_finite");
+    setLibcallName(RTLIB::LOG10_F128, "log10l");
+    setLibcallName(RTLIB::LOG10_FINITE_F128, "__log10l_finite");
+    setLibcallName(RTLIB::EXP_F128, "expl");
+    setLibcallName(RTLIB::EXP_FINITE_F128, "__expl_finite");
+    setLibcallName(RTLIB::EXP2_F128, "exp2l");
+    setLibcallName(RTLIB::EXP2_FINITE_F128, "__exp2l_finite");
+    setLibcallName(RTLIB::SIN_F128, "sinl");
+    setLibcallName(RTLIB::COS_F128, "cosl");
+    setLibcallName(RTLIB::POW_F128, "powl");
+    setLibcallName(RTLIB::POW_FINITE_F128, "__powl_finite");
+    setLibcallName(RTLIB::CEIL_F128, "ceill");
+    setLibcallName(RTLIB::TRUNC_F128, "truncl");
+    setLibcallName(RTLIB::RINT_F128, "rintl");
+    setLibcallName(RTLIB::NEARBYINT_F128, "nearbyintl");
+    setLibcallName(RTLIB::ROUND_F128, "roundl");
+    setLibcallName(RTLIB::ROUNDEVEN_F128, "roundevenl");
+    setLibcallName(RTLIB::FLOOR_F128, "floorl");
+    setLibcallName(RTLIB::COPYSIGN_F128, "copysignl");
+    setLibcallName(RTLIB::FMIN_F128, "fminl");
+    setLibcallName(RTLIB::FMAX_F128, "fmaxl");
+    setLibcallName(RTLIB::LROUND_F128, "lroundl");
+    setLibcallName(RTLIB::LLROUND_F128, "llroundl");
+    setLibcallName(RTLIB::LRINT_F128, "lrintl");
+    setLibcallName(RTLIB::LLRINT_F128, "llrintl");
+    setLibcallName(RTLIB::LDEXP_F128, "ldexpl");
+    setLibcallName(RTLIB::FREXP_F128, "frexpl");
+  }
 }
 
 /// GetFPLibCall - Helper to return the right libcall for the given floating



More information about the cfe-commits mailing list