llvmbot wrote:
<!--LLVM PR SUMMARY COMMENT--> @llvm/pr-subscribers-backend-arm Author: Matt Arsenault (arsenm) <details> <summary>Changes</summary> AArch64 was testing a couple of vector cases, and not the base scalars. Add the one case that isn't in the ARM version there, and then copy the rest back into AArch64. Also add a windows run line. --- Patch is 57.09 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/146691.diff 2 Files Affected: - (modified) llvm/test/CodeGen/AArch64/llvm.frexp.ll (+1345) - (modified) llvm/test/CodeGen/ARM/llvm.frexp.ll (+37) ``````````diff diff --git a/llvm/test/CodeGen/AArch64/llvm.frexp.ll b/llvm/test/CodeGen/AArch64/llvm.frexp.ll index e4cb8ed6eaf90..092fe16cbc2f3 100644 --- a/llvm/test/CodeGen/AArch64/llvm.frexp.ll +++ b/llvm/test/CodeGen/AArch64/llvm.frexp.ll @@ -1,5 +1,111 @@ ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 2 ; RUN: llc -mtriple=aarch64-gnu-linux < %s | FileCheck -check-prefixes=CHECK %s +; RUN: llc -mtriple=aarch64-windows-pc-msvc < %s | FileCheck -check-prefixes=WINDOWS %s +define { half, i32 } @test_frexp_f16_i32(half %a) { +; CHECK-LABEL: test_frexp_f16_i32: +; CHECK: // %bb.0: +; CHECK-NEXT: str x30, [sp, #-16]! // 8-byte Folded Spill +; CHECK-NEXT: .cfi_def_cfa_offset 16 +; CHECK-NEXT: .cfi_offset w30, -16 +; CHECK-NEXT: fcvt s0, h0 +; CHECK-NEXT: add x0, sp, #12 +; CHECK-NEXT: bl frexpf +; CHECK-NEXT: ldr w0, [sp, #12] +; CHECK-NEXT: fcvt h0, s0 +; CHECK-NEXT: ldr x30, [sp], #16 // 8-byte Folded Reload +; CHECK-NEXT: ret +; +; WINDOWS-LABEL: test_frexp_f16_i32: +; WINDOWS: .seh_proc test_frexp_f16_i32 +; WINDOWS-NEXT: // %bb.0: +; WINDOWS-NEXT: str x30, [sp, #-16]! // 8-byte Folded Spill +; WINDOWS-NEXT: .seh_save_reg_x x30, 16 +; WINDOWS-NEXT: .seh_endprologue +; WINDOWS-NEXT: fcvt d0, h0 +; WINDOWS-NEXT: add x0, sp, #12 +; WINDOWS-NEXT: bl frexp +; WINDOWS-NEXT: ldr w0, [sp, #12] +; WINDOWS-NEXT: fcvt h0, d0 +; WINDOWS-NEXT: .seh_startepilogue +; WINDOWS-NEXT: ldr x30, [sp], #16 // 8-byte Folded Reload +; WINDOWS-NEXT: .seh_save_reg_x x30, 16 +; WINDOWS-NEXT: .seh_endepilogue +; WINDOWS-NEXT: ret +; WINDOWS-NEXT: .seh_endfunclet +; WINDOWS-NEXT: .seh_endproc + %result = call { half, i32 } @llvm.frexp.f16.i32(half %a) + ret { half, i32 } %result +} + +define half @test_frexp_f16_i32_only_use_fract(half %a) { +; CHECK-LABEL: test_frexp_f16_i32_only_use_fract: +; CHECK: // %bb.0: +; CHECK-NEXT: str x30, [sp, #-16]! // 8-byte Folded Spill +; CHECK-NEXT: .cfi_def_cfa_offset 16 +; CHECK-NEXT: .cfi_offset w30, -16 +; CHECK-NEXT: fcvt s0, h0 +; CHECK-NEXT: add x0, sp, #12 +; CHECK-NEXT: bl frexpf +; CHECK-NEXT: fcvt h0, s0 +; CHECK-NEXT: ldr x30, [sp], #16 // 8-byte Folded Reload +; CHECK-NEXT: ret +; +; WINDOWS-LABEL: test_frexp_f16_i32_only_use_fract: +; WINDOWS: .seh_proc test_frexp_f16_i32_only_use_fract +; WINDOWS-NEXT: // %bb.0: +; WINDOWS-NEXT: str x30, [sp, #-16]! // 8-byte Folded Spill +; WINDOWS-NEXT: .seh_save_reg_x x30, 16 +; WINDOWS-NEXT: .seh_endprologue +; WINDOWS-NEXT: fcvt d0, h0 +; WINDOWS-NEXT: add x0, sp, #12 +; WINDOWS-NEXT: bl frexp +; WINDOWS-NEXT: fcvt h0, d0 +; WINDOWS-NEXT: .seh_startepilogue +; WINDOWS-NEXT: ldr x30, [sp], #16 // 8-byte Folded Reload +; WINDOWS-NEXT: .seh_save_reg_x x30, 16 +; WINDOWS-NEXT: .seh_endepilogue +; WINDOWS-NEXT: ret +; WINDOWS-NEXT: .seh_endfunclet +; WINDOWS-NEXT: .seh_endproc + %result = call { half, i32 } @llvm.frexp.f16.i32(half %a) + %result.0 = extractvalue { half, i32 } %result, 0 + ret half %result.0 +} + +define i32 @test_frexp_f16_i32_only_use_exp(half %a) { +; CHECK-LABEL: test_frexp_f16_i32_only_use_exp: +; CHECK: // %bb.0: +; CHECK-NEXT: str x30, [sp, #-16]! // 8-byte Folded Spill +; CHECK-NEXT: .cfi_def_cfa_offset 16 +; CHECK-NEXT: .cfi_offset w30, -16 +; CHECK-NEXT: fcvt s0, h0 +; CHECK-NEXT: add x0, sp, #12 +; CHECK-NEXT: bl frexpf +; CHECK-NEXT: ldr w0, [sp, #12] +; CHECK-NEXT: ldr x30, [sp], #16 // 8-byte Folded Reload +; CHECK-NEXT: ret +; +; WINDOWS-LABEL: test_frexp_f16_i32_only_use_exp: +; WINDOWS: .seh_proc test_frexp_f16_i32_only_use_exp +; WINDOWS-NEXT: // %bb.0: +; WINDOWS-NEXT: str x30, [sp, #-16]! // 8-byte Folded Spill +; WINDOWS-NEXT: .seh_save_reg_x x30, 16 +; WINDOWS-NEXT: .seh_endprologue +; WINDOWS-NEXT: fcvt d0, h0 +; WINDOWS-NEXT: add x0, sp, #12 +; WINDOWS-NEXT: bl frexp +; WINDOWS-NEXT: ldr w0, [sp, #12] +; WINDOWS-NEXT: .seh_startepilogue +; WINDOWS-NEXT: ldr x30, [sp], #16 // 8-byte Folded Reload +; WINDOWS-NEXT: .seh_save_reg_x x30, 16 +; WINDOWS-NEXT: .seh_endepilogue +; WINDOWS-NEXT: ret +; WINDOWS-NEXT: .seh_endfunclet +; WINDOWS-NEXT: .seh_endproc + %result = call { half, i32 } @llvm.frexp.f16.i32(half %a) + %result.0 = extractvalue { half, i32 } %result, 1 + ret i32 %result.0 +} define { <2 x half>, <2 x i32> } @test_frexp_v2f16_v2i32(<2 x half> %a) { ; CHECK-LABEL: test_frexp_v2f16_v2i32: @@ -51,10 +157,240 @@ define { <2 x half>, <2 x i32> } @test_frexp_v2f16_v2i32(<2 x half> %a) { ; CHECK-NEXT: // kill: def $d0 killed $d0 killed $q0 ; CHECK-NEXT: add sp, sp, #64 ; CHECK-NEXT: ret +; +; WINDOWS-LABEL: test_frexp_v2f16_v2i32: +; WINDOWS: .seh_proc test_frexp_v2f16_v2i32 +; WINDOWS-NEXT: // %bb.0: +; WINDOWS-NEXT: sub sp, sp, #64 +; WINDOWS-NEXT: .seh_stackalloc 64 +; WINDOWS-NEXT: str x19, [sp, #48] // 8-byte Folded Spill +; WINDOWS-NEXT: .seh_save_reg x19, 48 +; WINDOWS-NEXT: str x30, [sp, #56] // 8-byte Folded Spill +; WINDOWS-NEXT: .seh_save_reg x30, 56 +; WINDOWS-NEXT: .seh_endprologue +; WINDOWS-NEXT: // kill: def $d0 killed $d0 def $q0 +; WINDOWS-NEXT: mov h1, v0.h[1] +; WINDOWS-NEXT: str q0, [sp, #16] // 16-byte Folded Spill +; WINDOWS-NEXT: add x0, sp, #36 +; WINDOWS-NEXT: add x19, sp, #36 +; WINDOWS-NEXT: fcvt d0, h1 +; WINDOWS-NEXT: bl frexp +; WINDOWS-NEXT: ldr q1, [sp, #16] // 16-byte Folded Reload +; WINDOWS-NEXT: fcvt h0, d0 +; WINDOWS-NEXT: add x0, sp, #32 +; WINDOWS-NEXT: fcvt d1, h1 +; WINDOWS-NEXT: str q0, [sp] // 16-byte Folded Spill +; WINDOWS-NEXT: fmov d0, d1 +; WINDOWS-NEXT: bl frexp +; WINDOWS-NEXT: ldr q1, [sp, #16] // 16-byte Folded Reload +; WINDOWS-NEXT: fcvt h2, d0 +; WINDOWS-NEXT: add x0, sp, #40 +; WINDOWS-NEXT: mov h1, v1.h[2] +; WINDOWS-NEXT: fcvt d0, h1 +; WINDOWS-NEXT: ldr q1, [sp] // 16-byte Folded Reload +; WINDOWS-NEXT: mov v2.h[1], v1.h[0] +; WINDOWS-NEXT: str q2, [sp] // 16-byte Folded Spill +; WINDOWS-NEXT: bl frexp +; WINDOWS-NEXT: ldr q1, [sp, #16] // 16-byte Folded Reload +; WINDOWS-NEXT: fcvt h2, d0 +; WINDOWS-NEXT: add x0, sp, #44 +; WINDOWS-NEXT: mov h1, v1.h[3] +; WINDOWS-NEXT: fcvt d0, h1 +; WINDOWS-NEXT: ldr q1, [sp] // 16-byte Folded Reload +; WINDOWS-NEXT: mov v1.h[2], v2.h[0] +; WINDOWS-NEXT: str q1, [sp] // 16-byte Folded Spill +; WINDOWS-NEXT: bl frexp +; WINDOWS-NEXT: fcvt h2, d0 +; WINDOWS-NEXT: ldr s1, [sp, #32] +; WINDOWS-NEXT: ldr q0, [sp] // 16-byte Folded Reload +; WINDOWS-NEXT: ld1 { v1.s }[1], [x19] +; WINDOWS-NEXT: // kill: def $d1 killed $d1 killed $q1 +; WINDOWS-NEXT: mov v0.h[3], v2.h[0] +; WINDOWS-NEXT: // kill: def $d0 killed $d0 killed $q0 +; WINDOWS-NEXT: .seh_startepilogue +; WINDOWS-NEXT: ldr x30, [sp, #56] // 8-byte Folded Reload +; WINDOWS-NEXT: .seh_save_reg x30, 56 +; WINDOWS-NEXT: ldr x19, [sp, #48] // 8-byte Folded Reload +; WINDOWS-NEXT: .seh_save_reg x19, 48 +; WINDOWS-NEXT: add sp, sp, #64 +; WINDOWS-NEXT: .seh_stackalloc 64 +; WINDOWS-NEXT: .seh_endepilogue +; WINDOWS-NEXT: ret +; WINDOWS-NEXT: .seh_endfunclet +; WINDOWS-NEXT: .seh_endproc %result = call { <2 x half>, <2 x i32> } @llvm.frexp.v2f16.v2i32(<2 x half> %a) ret { <2 x half>, <2 x i32> } %result } +define <2 x half> @test_frexp_v2f16_v2i32_only_use_fract(<2 x half> %a) { +; CHECK-LABEL: test_frexp_v2f16_v2i32_only_use_fract: +; CHECK: // %bb.0: +; CHECK-NEXT: sub sp, sp, #64 +; CHECK-NEXT: str x30, [sp, #48] // 8-byte Folded Spill +; CHECK-NEXT: .cfi_def_cfa_offset 64 +; CHECK-NEXT: .cfi_offset w30, -16 +; CHECK-NEXT: // kill: def $d0 killed $d0 def $q0 +; CHECK-NEXT: mov h1, v0.h[1] +; CHECK-NEXT: str q0, [sp, #16] // 16-byte Folded Spill +; CHECK-NEXT: add x0, sp, #44 +; CHECK-NEXT: fcvt s0, h1 +; CHECK-NEXT: bl frexpf +; CHECK-NEXT: ldr q1, [sp, #16] // 16-byte Folded Reload +; CHECK-NEXT: fcvt h0, s0 +; CHECK-NEXT: add x0, sp, #40 +; CHECK-NEXT: fcvt s1, h1 +; CHECK-NEXT: str q0, [sp] // 16-byte Folded Spill +; CHECK-NEXT: fmov s0, s1 +; CHECK-NEXT: bl frexpf +; CHECK-NEXT: ldr q1, [sp, #16] // 16-byte Folded Reload +; CHECK-NEXT: fcvt h2, s0 +; CHECK-NEXT: add x0, sp, #56 +; CHECK-NEXT: mov h1, v1.h[2] +; CHECK-NEXT: fcvt s0, h1 +; CHECK-NEXT: ldr q1, [sp] // 16-byte Folded Reload +; CHECK-NEXT: mov v2.h[1], v1.h[0] +; CHECK-NEXT: str q2, [sp] // 16-byte Folded Spill +; CHECK-NEXT: bl frexpf +; CHECK-NEXT: ldr q1, [sp, #16] // 16-byte Folded Reload +; CHECK-NEXT: fcvt h2, s0 +; CHECK-NEXT: add x0, sp, #60 +; CHECK-NEXT: mov h1, v1.h[3] +; CHECK-NEXT: fcvt s0, h1 +; CHECK-NEXT: ldr q1, [sp] // 16-byte Folded Reload +; CHECK-NEXT: mov v1.h[2], v2.h[0] +; CHECK-NEXT: str q1, [sp] // 16-byte Folded Spill +; CHECK-NEXT: bl frexpf +; CHECK-NEXT: fcvt h1, s0 +; CHECK-NEXT: ldr q0, [sp] // 16-byte Folded Reload +; CHECK-NEXT: ldr x30, [sp, #48] // 8-byte Folded Reload +; CHECK-NEXT: mov v0.h[3], v1.h[0] +; CHECK-NEXT: // kill: def $d0 killed $d0 killed $q0 +; CHECK-NEXT: add sp, sp, #64 +; CHECK-NEXT: ret +; +; WINDOWS-LABEL: test_frexp_v2f16_v2i32_only_use_fract: +; WINDOWS: .seh_proc test_frexp_v2f16_v2i32_only_use_fract +; WINDOWS-NEXT: // %bb.0: +; WINDOWS-NEXT: sub sp, sp, #64 +; WINDOWS-NEXT: .seh_stackalloc 64 +; WINDOWS-NEXT: str x30, [sp, #48] // 8-byte Folded Spill +; WINDOWS-NEXT: .seh_save_reg x30, 48 +; WINDOWS-NEXT: .seh_endprologue +; WINDOWS-NEXT: // kill: def $d0 killed $d0 def $q0 +; WINDOWS-NEXT: mov h1, v0.h[1] +; WINDOWS-NEXT: str q0, [sp, #16] // 16-byte Folded Spill +; WINDOWS-NEXT: add x0, sp, #44 +; WINDOWS-NEXT: fcvt d0, h1 +; WINDOWS-NEXT: bl frexp +; WINDOWS-NEXT: ldr q1, [sp, #16] // 16-byte Folded Reload +; WINDOWS-NEXT: fcvt h0, d0 +; WINDOWS-NEXT: add x0, sp, #40 +; WINDOWS-NEXT: fcvt d1, h1 +; WINDOWS-NEXT: str q0, [sp] // 16-byte Folded Spill +; WINDOWS-NEXT: fmov d0, d1 +; WINDOWS-NEXT: bl frexp +; WINDOWS-NEXT: ldr q1, [sp, #16] // 16-byte Folded Reload +; WINDOWS-NEXT: fcvt h2, d0 +; WINDOWS-NEXT: add x0, sp, #56 +; WINDOWS-NEXT: mov h1, v1.h[2] +; WINDOWS-NEXT: fcvt d0, h1 +; WINDOWS-NEXT: ldr q1, [sp] // 16-byte Folded Reload +; WINDOWS-NEXT: mov v2.h[1], v1.h[0] +; WINDOWS-NEXT: str q2, [sp] // 16-byte Folded Spill +; WINDOWS-NEXT: bl frexp +; WINDOWS-NEXT: ldr q1, [sp, #16] // 16-byte Folded Reload +; WINDOWS-NEXT: fcvt h2, d0 +; WINDOWS-NEXT: add x0, sp, #60 +; WINDOWS-NEXT: mov h1, v1.h[3] +; WINDOWS-NEXT: fcvt d0, h1 +; WINDOWS-NEXT: ldr q1, [sp] // 16-byte Folded Reload +; WINDOWS-NEXT: mov v1.h[2], v2.h[0] +; WINDOWS-NEXT: str q1, [sp] // 16-byte Folded Spill +; WINDOWS-NEXT: bl frexp +; WINDOWS-NEXT: fcvt h1, d0 +; WINDOWS-NEXT: ldr q0, [sp] // 16-byte Folded Reload +; WINDOWS-NEXT: mov v0.h[3], v1.h[0] +; WINDOWS-NEXT: // kill: def $d0 killed $d0 killed $q0 +; WINDOWS-NEXT: .seh_startepilogue +; WINDOWS-NEXT: ldr x30, [sp, #48] // 8-byte Folded Reload +; WINDOWS-NEXT: .seh_save_reg x30, 48 +; WINDOWS-NEXT: add sp, sp, #64 +; WINDOWS-NEXT: .seh_stackalloc 64 +; WINDOWS-NEXT: .seh_endepilogue +; WINDOWS-NEXT: ret +; WINDOWS-NEXT: .seh_endfunclet +; WINDOWS-NEXT: .seh_endproc + %result = call { <2 x half>, <2 x i32> } @llvm.frexp.v2f16.v2i32(<2 x half> %a) + %result.0 = extractvalue { <2 x half>, <2 x i32> } %result, 0 + ret <2 x half> %result.0 +} + +define <2 x i32> @test_frexp_v2f16_v2i32_only_use_exp(<2 x half> %a) { +; CHECK-LABEL: test_frexp_v2f16_v2i32_only_use_exp: +; CHECK: // %bb.0: +; CHECK-NEXT: sub sp, sp, #48 +; CHECK-NEXT: stp x30, x19, [sp, #32] // 16-byte Folded Spill +; CHECK-NEXT: .cfi_def_cfa_offset 48 +; CHECK-NEXT: .cfi_offset w19, -8 +; CHECK-NEXT: .cfi_offset w30, -16 +; CHECK-NEXT: // kill: def $d0 killed $d0 def $q0 +; CHECK-NEXT: str q0, [sp] // 16-byte Folded Spill +; CHECK-NEXT: add x0, sp, #24 +; CHECK-NEXT: fcvt s0, h0 +; CHECK-NEXT: bl frexpf +; CHECK-NEXT: ldr q0, [sp] // 16-byte Folded Reload +; CHECK-NEXT: add x0, sp, #28 +; CHECK-NEXT: add x19, sp, #28 +; CHECK-NEXT: mov h0, v0.h[1] +; CHECK-NEXT: fcvt s0, h0 +; CHECK-NEXT: bl frexpf +; CHECK-NEXT: ldr s0, [sp, #24] +; CHECK-NEXT: ld1 { v0.s }[1], [x19] +; CHECK-NEXT: ldp x30, x19, [sp, #32] // 16-byte Folded Reload +; CHECK-NEXT: // kill: def $d0 killed $d0 killed $q0 +; CHECK-NEXT: add sp, sp, #48 +; CHECK-NEXT: ret +; +; WINDOWS-LABEL: test_frexp_v2f16_v2i32_only_use_exp: +; WINDOWS: .seh_proc test_frexp_v2f16_v2i32_only_use_exp +; WINDOWS-NEXT: // %bb.0: +; WINDOWS-NEXT: sub sp, sp, #48 +; WINDOWS-NEXT: .seh_stackalloc 48 +; WINDOWS-NEXT: str x19, [sp, #32] // 8-byte Folded Spill +; WINDOWS-NEXT: .seh_save_reg x19, 32 +; WINDOWS-NEXT: str x30, [sp, #40] // 8-byte Folded Spill +; WINDOWS-NEXT: .seh_save_reg x30, 40 +; WINDOWS-NEXT: .seh_endprologue +; WINDOWS-NEXT: // kill: def $d0 killed $d0 def $q0 +; WINDOWS-NEXT: str q0, [sp] // 16-byte Folded Spill +; WINDOWS-NEXT: add x0, sp, #24 +; WINDOWS-NEXT: fcvt d0, h0 +; WINDOWS-NEXT: bl frexp +; WINDOWS-NEXT: ldr q0, [sp] // 16-byte Folded Reload +; WINDOWS-NEXT: add x0, sp, #28 +; WINDOWS-NEXT: add x19, sp, #28 +; WINDOWS-NEXT: mov h0, v0.h[1] +; WINDOWS-NEXT: fcvt d0, h0 +; WINDOWS-NEXT: bl frexp +; WINDOWS-NEXT: ldr s0, [sp, #24] +; WINDOWS-NEXT: ld1 { v0.s }[1], [x19] +; WINDOWS-NEXT: // kill: def $d0 killed $d0 killed $q0 +; WINDOWS-NEXT: .seh_startepilogue +; WINDOWS-NEXT: ldr x30, [sp, #40] // 8-byte Folded Reload +; WINDOWS-NEXT: .seh_save_reg x30, 40 +; WINDOWS-NEXT: ldr x19, [sp, #32] // 8-byte Folded Reload +; WINDOWS-NEXT: .seh_save_reg x19, 32 +; WINDOWS-NEXT: add sp, sp, #48 +; WINDOWS-NEXT: .seh_stackalloc 48 +; WINDOWS-NEXT: .seh_endepilogue +; WINDOWS-NEXT: ret +; WINDOWS-NEXT: .seh_endfunclet +; WINDOWS-NEXT: .seh_endproc + %result = call { <2 x half>, <2 x i32> } @llvm.frexp.v2f16.v2i32(<2 x half> %a) + %result.1 = extractvalue { <2 x half>, <2 x i32> } %result, 1 + ret <2 x i32> %result.1 +} + define { <3 x float>, <3 x i32> } @test_frexp_v3f16_v3i32(<3 x float> %a) { ; CHECK-LABEL: test_frexp_v3f16_v3i32: ; CHECK: // %bb.0: @@ -96,6 +432,1015 @@ define { <3 x float>, <3 x i32> } @test_frexp_v3f16_v3i32(<3 x float> %a) { ; CHECK-NEXT: mov v0.16b, v2.16b ; CHECK-NEXT: add sp, sp, #80 ; CHECK-NEXT: ret +; +; WINDOWS-LABEL: test_frexp_v3f16_v3i32: +; WINDOWS: .seh_proc test_frexp_v3f16_v3i32 +; WINDOWS-NEXT: // %bb.0: +; WINDOWS-NEXT: sub sp, sp, #80 +; WINDOWS-NEXT: .seh_stackalloc 80 +; WINDOWS-NEXT: stp x19, x20, [sp, #48] // 16-byte Folded Spill +; WINDOWS-NEXT: .seh_save_regp x19, 48 +; WINDOWS-NEXT: stp x21, x30, [sp, #64] // 16-byte Folded Spill +; WINDOWS-NEXT: .seh_save_lrpair x21, 64 +; WINDOWS-NEXT: .seh_endprologue +; WINDOWS-NEXT: mov s1, v0.s[1] +; WINDOWS-NEXT: str q0, [sp, #16] // 16-byte Folded Spill +; WINDOWS-NEXT: add x0, sp, #36 +; WINDOWS-NEXT: add x19, sp, #36 +; WINDOWS-NEXT: fcvt d0, s1 +; WINDOWS-NEXT: bl frexp +; WINDOWS-NEXT: fcvt s0, d0 +; WINDOWS-NEXT: add x0, sp, #32 +; WINDOWS-NEXT: str q0, [sp] // 16-byte Folded Spill +; WINDOWS-NEXT: ldr q0, [sp, #16] // 16-byte Folded Reload +; WINDOWS-NEXT: fcvt d0, s0 +; WINDOWS-NEXT: bl frexp +; WINDOWS-NEXT: fcvt s0, d0 +; WINDOWS-NEXT: ldr q1, [sp] // 16-byte Folded Reload +; WINDOWS-NEXT: add x0, sp, #40 +; WINDOWS-NEXT: add x20, sp, #40 +; WINDOWS-NEXT: mov v0.s[1], v1.s[0] +; WINDOWS-NEXT: str q0, [sp] // 16-byte Folded Spill +; WINDOWS-NEXT: ldr q0, [sp, #16] // 16-byte Folded Reload +; WINDOWS-NEXT: mov s0, v0.s[2] +; WINDOWS-NEXT: fcvt d0, s0 +; WINDOWS-NEXT: bl frexp +; WINDOWS-NEXT: fcvt s0, d0 +; WINDOWS-NEXT: ldr q1, [sp] // 16-byte Folded Reload +; WINDOWS-NEXT: add x0, sp, #44 +; WINDOWS-NEXT: add x21, sp, #44 +; WINDOWS-NEXT: mov v1.s[2], v0.s[0] +; WINDOWS-NEXT: ldr q0, [sp, #16] // 16-byte Folded Reload +; WINDOWS-NEXT: mov s0, v0.s[3] +; WINDOWS-NEXT: str q1, [sp] // 16-byte Folded Spill +; WINDOWS-NEXT: fcvt d0, s0 +; WINDOWS-NEXT: bl frexp +; WINDOWS-NEXT: ldr s1, [sp, #32] +; WINDOWS-NEXT: fcvt s2, d0 +; WINDOWS-NEXT: ldr q0, [sp] // 16-byte Folded Reload +; WINDOWS-NEXT: ld1 { v1.s }[1], [x19] +; WINDOWS-NEXT: mov v0.s[3], v2.s[0] +; WINDOWS-NEXT: ld1 { v1.s }[2], [x20] +; WINDOWS-NEXT: ld1 { v1.s }[3], [x21] +; WINDOWS-NEXT: .seh_startepilogue +; WINDOWS-NEXT: ldp x21, x30, [sp, #64] // 16-byte Folded Reload +; WINDOWS-NEXT: .seh_save_lrpair x21, 64 +; WINDOWS-NEXT: ldp x19, x20, [sp, #48] // 16-byte Folded Reload +; WINDOWS-NEXT: .seh_save_regp x19, 48 +; WINDOWS-NEXT: add sp, sp, #80 +; WINDOWS-NEXT: .seh_stackalloc 80 +; WINDOWS-NEXT: .seh_endepilogue +; WINDOWS-NEXT: ret +; WINDOWS-NEXT: .seh_endfunclet +; WINDOWS-NEXT: .seh_endproc %result = call { <3 x float>, <3 x i32> } @llvm.frexp.v3float.v3i32(<3 x float> %a) ret { <3 x float>, <3 x i32> } %result } + +define { float, i32 } @test_frexp_f32_i32(float %a) { +; CHECK-LABEL: test_frexp_f32_i32: +; CHECK: // %bb.0: +; CHECK-NEXT: str x30, [sp, #-16]! // 8-byte Folded Spill +; CHECK-NEXT: .cfi_def_cfa_offset 16 +; CHECK-NEXT: .cfi_offset w30, -16 +; CHECK-NEXT: add x0, sp, #12 +; CHECK-NEXT: bl frexpf +; CHECK-NEXT: ldr w0, [sp, #12] +; CHECK-NEXT: ldr x30, [sp], #16 // 8-byte Folded Reload +; CHECK-NEXT: ret +; +; WINDOWS-LABEL: test_frexp_f32_i32: +; WINDOWS: .seh_proc test_frexp_f32_i32 +; WINDOWS-NEXT: // %bb.0: +; WINDOWS-NEXT: str x30, [sp, #-16]! // 8-byte Folded Spill +; WINDOWS-NEXT: .seh_save_reg_x x30, 16 +; WINDOWS-NEXT: .seh_endprologue +; WINDOWS-NEXT: fcvt d0, s0 +; WINDOWS-NEXT: add x0, sp, #12 +; WINDOWS-NEXT: bl frexp +; WINDOWS-NEXT: ldr w0, [sp, #12] +; WINDOWS-NEXT: fcvt s0, d0 +; WINDOWS-NEXT: .seh_startepilogue +; WINDOWS-NEXT: ldr x30, [sp], #16 // 8-byte Folded Reload +; WINDOWS-NEXT: .seh_save_reg_x x30, 16 +; WINDOWS-NEXT: .seh_endepilogue +; WINDOWS-NEXT: ret +; WINDOWS-NEXT: .seh_endfunclet +; WINDOWS-NEXT: .seh_endproc + %result = call { float, i32 } @llvm.frexp.f32.i32(float %a) + ret { float, i32 } %result +} + +define { float, i32 } @test_frexp_f32_i32_tailcall(float %a) { +; CHECK-LABEL: test_frexp_f32_i32_tailcall: +; CHECK: // %bb.0: +; CHECK-NEXT: str x30, [sp, #-16]! // 8-byte Folded Spill +; CHECK-NEXT: .cfi_def_cfa_offset 16 +; CHECK-NEXT: .cfi_offset w30, -16 +; CHECK-NEXT: add x0, sp, #12 +; CHECK-NEXT: bl frexpf +; CHECK-NEXT: ldr w0, [sp, #12] +; CHECK-NEXT: ldr x30, [sp], #16 // 8-byte Folded Reload +; CHECK-NEXT: ret +; +; WINDOWS-LABEL: test_frexp_f32_i32_tailcall: +; WINDOWS: .seh_proc test_frexp_f32_i32_tailcall +; WINDOWS-NEXT: // %bb.0: +; WINDOWS-NEXT: str x30, [sp, #-16]! // 8-byte Folded Spill +; WINDOWS-NEXT: .seh_save_reg_x x30, 16 +; WINDOWS-NEXT: .seh_endprologue +; WINDOWS-NEXT: fcvt d0, s0 +; WINDOWS-NEXT: add x0, sp, #12 +; WINDOWS-NEXT: bl frexp +; WINDOWS-NEXT: ldr w0, [sp, #12] +; WINDOWS-NEXT: fcvt s0, d0 +; WINDOWS-NEXT: .seh_startepilogue +; WINDOWS-NEXT: ldr x30, [sp], #16 // 8-byte Folded Reload +; WINDOWS-NEXT: .seh_save_reg_x x30, 16 +; WINDOWS-NEXT: .seh_endepilogue +; WINDOWS-NEXT: ret +; WINDOWS-NEXT: .seh_endfunclet +; WINDOWS-NEXT: .seh_endproc + %result = tail call { float, i32 } @llvm.frexp.f32.i32(float %a) + ret { float, i32 } %result +} + +define float @test_frexp_f32_i32_only_use_fract(float %a) { +; CHECK-LABEL: test_frexp_f32_i32_only_use_fract: +; CHECK: // %bb.0: +; CHECK-NEXT: str x30, [sp, #-16]! // 8-byte Folded Spill +; CHECK-NEXT: .cfi_def_cfa_offset 16 +; CHECK-NEXT: .cfi_offset w30, -16 +; CHECK-NEXT: add x0, sp, #12 +; CHECK-NEXT: bl frexpf +; CHECK-NEXT: ldr x30, [sp], #16 // 8-byte Folded Reload +; CHECK-NEXT: ret +; +; WINDOWS-LABEL: test_frexp_f32_i32_only_use_... [truncated] `````````` </details> https://github.com/llvm/llvm-project/pull/146691 _______________________________________________ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
