https://github.com/c-rhodes created https://github.com/llvm/llvm-project/pull/231406
This was disabled earlier this year (#177359) before being shortly reverted due to two test failures which have since been fixed. The flaky LLDB issue was fixed in #210050 and the libcxxabi test in #230518. CTMark aarch64-O0-g geomean improves -0.97%. Code size regresses 2.70% but as previously we think this is worth paying at O0 where compile-time and debug are the biggest concern. https://llvm-compile-time-tracker.com/compare.php?from=fc90bd3cfb0bab14a2bf7cbd33f5f1e904c7e133&to=071f3542ccb3b3c65032397550e862d7090f425f&stat=instructions:u Assisted-by: codex Co-authored-by: David Green (davemgreen) >From 692fd15b1e9b4b237e377c3b86aef75c7bb22350 Mon Sep 17 00:00:00 2001 From: Cullen Rhodes <[email protected]> Date: Mon, 12 Oct 2026 14:36:34 +0000 Subject: [PATCH] [AArch64][GlobalISel] Do not run the Localizer at -O0 --- bolt/test/AArch64/icp-inline.c | 4 +- bolt/test/AArch64/jmp-table-unsupported.s | 2 +- .../AArch64/instrumentation-ind-call.c | 4 +- clang/test/CodeGen/memtag-globals-asm.cpp | 25 +-- .../Target/AArch64/AArch64TargetMachine.cpp | 1 - .../AArch64/GlobalISel/arm64-atomic.ll | 55 +++--- .../atomic-anyextending-load-crash.ll | 4 +- .../AArch64/GlobalISel/invoke-region.ll | 14 +- .../AArch64/GlobalISel/legalize-exceptions.ll | 9 +- .../AArch64/GlobalISel/localizer-arm64-tti.ll | 98 ++++------- .../localizer-propagate-debug-loc.mir | 37 ++-- .../AArch64/GlobalISel/localizer-tls.ll | 16 +- .../GlobalISel/select-fp-anyext-crash.ll | 22 +-- .../GlobalISel/translate-constant-dag.ll | 4 +- llvm/test/CodeGen/AArch64/O0-pipeline.ll | 1 - llvm/test/CodeGen/AArch64/O3-pipeline.ll | 1 - .../CodeGen/AArch64/aarch64-fastcc-stackup.ll | 2 +- llvm/test/CodeGen/AArch64/aarch64-tbz.ll | 12 +- llvm/test/CodeGen/AArch64/cfguard-checks.ll | 4 +- llvm/test/CodeGen/AArch64/cpa-globalisel.ll | 166 +++++++++--------- llvm/test/CodeGen/AArch64/large-stack.ll | 15 +- llvm/test/CodeGen/AArch64/phi-dbg.ll | 3 +- llvm/test/CodeGen/AArch64/popcount.ll | 38 ++-- llvm/test/CodeGen/AArch64/pr92062.ll | 2 +- llvm/test/CodeGen/AArch64/vararg.ll | 16 +- .../CodeGen/AArch64/win64_vararg_float.ll | 12 +- .../CodeGen/AArch64/win64_vararg_float_cc.ll | 12 +- llvm/test/Other/print-changed-machine.ll | 2 +- 28 files changed, 289 insertions(+), 292 deletions(-) diff --git a/bolt/test/AArch64/icp-inline.c b/bolt/test/AArch64/icp-inline.c index f2a4f47147eaa..97c407223af91 100644 --- a/bolt/test/AArch64/icp-inline.c +++ b/bolt/test/AArch64/icp-inline.c @@ -30,8 +30,8 @@ __attribute__((noinline)) int indirectTailCall(int Argc) { __attribute__((musttail)) return Func(0); } -// FDATA: 1 indirectTailCall 28 1 foo 0 0 1 -// FDATA: 1 indirectTailCall 28 1 bar 0 0 2 +// FDATA: 1 indirectTailCall 2c 1 foo 0 0 1 +// FDATA: 1 indirectTailCall 2c 1 bar 0 0 2 int main(int Argc, char **Argv) { (void)Argv; diff --git a/bolt/test/AArch64/jmp-table-unsupported.s b/bolt/test/AArch64/jmp-table-unsupported.s index 1228149430449..ad0f7cef838af 100644 --- a/bolt/test/AArch64/jmp-table-unsupported.s +++ b/bolt/test/AArch64/jmp-table-unsupported.s @@ -101,7 +101,7 @@ SECTIONS { # JT-OBJDUMP-NORMAL: <handleOptionJumpTable>: # JT-OBJDUMP-NORMAL: adrp # JT-OBJDUMP-NORMAL-NEXT: add -# JT-OBJDUMP-NORMAL-NEXT: ldr +# JT-OBJDUMP-NORMAL: ldr # JT-OBJDUMP-NORMAL-NEXT: blr # RUN: llvm-bolt %t/jt_type_normal.exe --dyno-stats \ diff --git a/bolt/test/runtime/AArch64/instrumentation-ind-call.c b/bolt/test/runtime/AArch64/instrumentation-ind-call.c index 8d9d1e13060f5..a0c1eed7e28bd 100644 --- a/bolt/test/runtime/AArch64/instrumentation-ind-call.c +++ b/bolt/test/runtime/AArch64/instrumentation-ind-call.c @@ -35,7 +35,7 @@ RUN: | FileCheck %s --check-prefix=CHECKINDIRECTREG CHECKINDIRECTREG: mov w0, #0xa CHECKINDIRECTREG-NEXT: mov w1, #0x14 -CHECKINDIRECTREG-NEXT: blr x8 +CHECKINDIRECTREG: blr x8 RUN: llvm-bolt %t.exe --instrument --instrumentation-file=%t.fdata \ RUN: -o %t.instrumented \ @@ -57,7 +57,7 @@ RUN: %t.instrumented | FileCheck %s --check-prefix=CHECK-INSTR-INDIR-CALL-FUNC CHECK-MAIN: mov w0, #0xa CHECK-MAIN-NEXT: mov w1, #0x14 // store current values -CHECK-MAIN-NEXT: stp x0, x30, [sp +CHECK-MAIN: stp x0, x30, [sp // load callsite id CHECK-MAIN-NEXT: mov x0, CHECK-MAIN-NEXT: stp x8, x0, [sp diff --git a/clang/test/CodeGen/memtag-globals-asm.cpp b/clang/test/CodeGen/memtag-globals-asm.cpp index fb3958dd8bcb6..331939e49ac9d 100644 --- a/clang/test/CodeGen/memtag-globals-asm.cpp +++ b/clang/test/CodeGen/memtag-globals-asm.cpp @@ -248,32 +248,33 @@ int f(int x) { global_my_class.virtual_func() + local_my_class.virtual_func() + local_const_string[0] + + // GOT pointers may be spilled before use at -O0. // CHECK-Q-DAG: adrp [[REG_J:x[0-9]+]], :got:_ZL12local_string - // CHECK-Q-DAG: ldr [[REG_J2:x[0-9]+]], [[[REG_J]], :got_lo12:_ZL12local_string] - // CHECK-Q-DAG: ldrsb {{.*}}, [[[REG_J2]]] + // CHECK-Q-DAG: ldr {{x[0-9]+}}, [[[REG_J]], :got_lo12:_ZL12local_string] + // CHECK-Q-DAG: ldrsb {{.*}}, [{{x[0-9]+}}] local_string[0] + // CHECK-Q-DAG: adrp [[REG_K:x[0-9]+]], :got:_ZL16local_atomic_int - // CHECK-Q-DAG: ldr [[REG_K2:x[0-9]+]], [[[REG_K]], :got_lo12:_ZL16local_atomic_int] - // CHECK-Q-DAG: ldar {{.*}}, [[[REG_K2]]] + // CHECK-Q-DAG: ldr {{x[0-9]+}}, [[[REG_K]], :got_lo12:_ZL16local_atomic_int] + // CHECK-Q-DAG: ldar {{.*}}, [{{x[0-9]+}}] local_atomic_int + // CHECK-Q-DAG: adrp [[REG_L:x[0-9]+]], :got:global_atomic_int - // CHECK-Q-DAG: ldr [[REG_L2:x[0-9]+]], [[[REG_L]], :got_lo12:global_atomic_int] - // CHECK-Q-DAG: ldar {{.*}}, [[[REG_L2]]] + // CHECK-Q-DAG: ldr {{x[0-9]+}}, [[[REG_L]], :got_lo12:global_atomic_int] + // CHECK-Q-DAG: ldar {{.*}}, [{{x[0-9]+}}] global_atomic_int + // CHECK-Q-DAG: adrp [[REG_M:x[0-9]+]], :got:global_union - // CHECK-Q-DAG: ldr [[REG_M2:x[0-9]+]], [[[REG_M]], :got_lo12:global_union] - // CHECK-Q-DAG: ldr {{.*}}, [[[REG_M2]]] + // CHECK-Q-DAG: ldr {{x[0-9]+}}, [[[REG_M]], :got_lo12:global_union] + // CHECK-Q-DAG: ldr {{.*}}, [{{x[0-9]+}}] global_union.i + // CHECK-Q-DAG: adrp [[REG_N:x[0-9]+]], :got:_ZL11local_union - // CHECK-Q-DAG: ldr [[REG_N2:x[0-9]+]], [[[REG_N]], :got_lo12:_ZL11local_union] - // CHECK-Q-DAG: ldrsb {{.*}}, [[[REG_N2]]] + // CHECK-Q-DAG: ldr {{x[0-9]+}}, [[[REG_N]], :got_lo12:_ZL11local_union] + // CHECK-Q-DAG: ldrsb {{.*}}, [{{x[0-9]+}}] local_union.c + /// Global variables - ignore. global_tls + local_tls + // CHECK-Q-DAG: adrp [[REG_O:x[0-9]+]], :got:_ZZ1fiE12function_int - // CHECK-Q-DAG: ldr [[REG_O2:x[0-9]+]], [[[REG_O]], :got_lo12:_ZZ1fiE12function_int] - // CHECK-Q-DAG: ldr {{.*}}, [[[REG_O2]]] + // CHECK-Q-DAG: ldr {{x[0-9]+}}, [[[REG_O]], :got_lo12:_ZZ1fiE12function_int] + // CHECK-Q-DAG: ldr {{.*}}, [{{x[0-9]+}}] function_int; } diff --git a/llvm/lib/Target/AArch64/AArch64TargetMachine.cpp b/llvm/lib/Target/AArch64/AArch64TargetMachine.cpp index 59fb0e15649b8..5741535fe783b 100644 --- a/llvm/lib/Target/AArch64/AArch64TargetMachine.cpp +++ b/llvm/lib/Target/AArch64/AArch64TargetMachine.cpp @@ -618,7 +618,6 @@ bool AArch64PassConfig::addIRTranslator() { void AArch64PassConfig::addPreLegalizeMachineIR() { if (getAArch64TargetMachine().isGlobalISelOptNone()) { addPass(createAArch64O0PreLegalizerCombiner()); - addPass(new LocalizerLegacy()); } else { addPass(createAArch64PreLegalizerCombiner()); addPass(new LocalizerLegacy()); diff --git a/llvm/test/CodeGen/AArch64/GlobalISel/arm64-atomic.ll b/llvm/test/CodeGen/AArch64/GlobalISel/arm64-atomic.ll index 6da7795f8b5e8..56925d384995f 100644 --- a/llvm/test/CodeGen/AArch64/GlobalISel/arm64-atomic.ll +++ b/llvm/test/CodeGen/AArch64/GlobalISel/arm64-atomic.ll @@ -762,6 +762,8 @@ define i32 @fetch_and_or(ptr %p) #0 { ; CHECK-NOLSE-O0: ; %bb.0: ; CHECK-NOLSE-O0-NEXT: sub sp, sp, #32 ; CHECK-NOLSE-O0-NEXT: str x0, [sp, #16] ; 8-byte Spill +; CHECK-NOLSE-O0-NEXT: mov w8, #5 ; =0x5 +; CHECK-NOLSE-O0-NEXT: str w8, [sp, #24] ; 4-byte Spill ; CHECK-NOLSE-O0-NEXT: ldr w8, [x0] ; CHECK-NOLSE-O0-NEXT: str w8, [sp, #28] ; 4-byte Spill ; CHECK-NOLSE-O0-NEXT: b LBB8_1 @@ -770,7 +772,7 @@ define i32 @fetch_and_or(ptr %p) #0 { ; CHECK-NOLSE-O0-NEXT: ; Child Loop BB8_2 Depth 2 ; CHECK-NOLSE-O0-NEXT: ldr w10, [sp, #28] ; 4-byte Reload ; CHECK-NOLSE-O0-NEXT: ldr x11, [sp, #16] ; 8-byte Reload -; CHECK-NOLSE-O0-NEXT: mov w8, #5 ; =0x5 +; CHECK-NOLSE-O0-NEXT: ldr w8, [sp, #24] ; 4-byte Reload ; CHECK-NOLSE-O0-NEXT: orr w12, w10, w8 ; CHECK-NOLSE-O0-NEXT: LBB8_2: ; %atomicrmw.start ; CHECK-NOLSE-O0-NEXT: ; Parent Loop BB8_1 Depth=1 @@ -1829,8 +1831,8 @@ define void @atomic_truncstore_relaxed_32(ptr %p, i32 %off32, i64 %inp1, i64 %in ; CHECK-NOLSE-O0-NEXT: str w8, [x0, w1, sxtw #2] ; CHECK-NOLSE-O0-NEXT: mov w8, w4 ; CHECK-NOLSE-O0-NEXT: stur w8, [x0, #-256] -; CHECK-NOLSE-O0-NEXT: mov w8, w5 ; CHECK-NOLSE-O0-NEXT: add x9, x0, #291, lsl #12 ; =1191936 +; CHECK-NOLSE-O0-NEXT: mov w8, w5 ; CHECK-NOLSE-O0-NEXT: str w8, [x9] ; CHECK-NOLSE-O0-NEXT: ret ; @@ -1842,8 +1844,8 @@ define void @atomic_truncstore_relaxed_32(ptr %p, i32 %off32, i64 %inp1, i64 %in ; CHECK-OUTLINE-O0-NEXT: str w8, [x0, w1, sxtw #2] ; CHECK-OUTLINE-O0-NEXT: mov w8, w4 ; CHECK-OUTLINE-O0-NEXT: stur w8, [x0, #-256] -; CHECK-OUTLINE-O0-NEXT: mov w8, w5 ; CHECK-OUTLINE-O0-NEXT: add x9, x0, #291, lsl #12 ; =1191936 +; CHECK-OUTLINE-O0-NEXT: mov w8, w5 ; CHECK-OUTLINE-O0-NEXT: str w8, [x9] ; CHECK-OUTLINE-O0-NEXT: ret ; @@ -1864,8 +1866,8 @@ define void @atomic_truncstore_relaxed_32(ptr %p, i32 %off32, i64 %inp1, i64 %in ; CHECK-LSE-O0-NEXT: str w8, [x0, w1, sxtw #2] ; CHECK-LSE-O0-NEXT: mov x8, x4 ; CHECK-LSE-O0-NEXT: stur w8, [x0, #-256] -; CHECK-LSE-O0-NEXT: mov x8, x5 ; CHECK-LSE-O0-NEXT: add x9, x0, #291, lsl #12 ; =1191936 +; CHECK-LSE-O0-NEXT: mov x8, x5 ; CHECK-LSE-O0-NEXT: str w8, [x9] ; CHECK-LSE-O0-NEXT: ret %ptr_unsigned = getelementptr i32, ptr %p, i32 4095 @@ -2081,14 +2083,17 @@ define void @store_trunc(i32 %val, ptr %p8, ptr %p16, i64 %inp64) { ; CHECK-NOLSE-O0-NEXT: .cfi_def_cfa_offset 32 ; CHECK-NOLSE-O0-NEXT: .cfi_offset w30, -8 ; CHECK-NOLSE-O0-NEXT: .cfi_offset w29, -16 -; CHECK-NOLSE-O0-NEXT: stlrb w0, [x1] -; CHECK-NOLSE-O0-NEXT: strh w0, [x2] +; CHECK-NOLSE-O0-NEXT: mov w9, w0 +; CHECK-NOLSE-O0-NEXT: mov x10, x2 +; CHECK-NOLSE-O0-NEXT: mov x8, x3 +; CHECK-NOLSE-O0-NEXT: mov w11, #4 ; =0x4 +; CHECK-NOLSE-O0-NEXT: mov w0, w11 +; CHECK-NOLSE-O0-NEXT: mov w3, #5 ; =0x5 ; CHECK-NOLSE-O0-NEXT: add x2, sp, #12 -; CHECK-NOLSE-O0-NEXT: mov w8, w3 +; CHECK-NOLSE-O0-NEXT: stlrb w9, [x1] +; CHECK-NOLSE-O0-NEXT: strh w9, [x10] +; CHECK-NOLSE-O0-NEXT: ; kill: def $w8 killed $w8 killed $x8 ; CHECK-NOLSE-O0-NEXT: str w8, [sp, #12] -; CHECK-NOLSE-O0-NEXT: mov w8, #4 ; =0x4 -; CHECK-NOLSE-O0-NEXT: mov w0, w8 -; CHECK-NOLSE-O0-NEXT: mov w3, #5 ; =0x5 ; CHECK-NOLSE-O0-NEXT: bl ___atomic_store ; CHECK-NOLSE-O0-NEXT: ldp x29, x30, [sp, #16] ; 16-byte Folded Reload ; CHECK-NOLSE-O0-NEXT: add sp, sp, #32 @@ -2101,14 +2106,17 @@ define void @store_trunc(i32 %val, ptr %p8, ptr %p16, i64 %inp64) { ; CHECK-OUTLINE-O0-NEXT: .cfi_def_cfa_offset 32 ; CHECK-OUTLINE-O0-NEXT: .cfi_offset w30, -8 ; CHECK-OUTLINE-O0-NEXT: .cfi_offset w29, -16 -; CHECK-OUTLINE-O0-NEXT: stlrb w0, [x1] -; CHECK-OUTLINE-O0-NEXT: strh w0, [x2] +; CHECK-OUTLINE-O0-NEXT: mov w9, w0 +; CHECK-OUTLINE-O0-NEXT: mov x10, x2 +; CHECK-OUTLINE-O0-NEXT: mov x8, x3 +; CHECK-OUTLINE-O0-NEXT: mov w11, #4 ; =0x4 +; CHECK-OUTLINE-O0-NEXT: mov w0, w11 +; CHECK-OUTLINE-O0-NEXT: mov w3, #5 ; =0x5 ; CHECK-OUTLINE-O0-NEXT: add x2, sp, #12 -; CHECK-OUTLINE-O0-NEXT: mov w8, w3 +; CHECK-OUTLINE-O0-NEXT: stlrb w9, [x1] +; CHECK-OUTLINE-O0-NEXT: strh w9, [x10] +; CHECK-OUTLINE-O0-NEXT: ; kill: def $w8 killed $w8 killed $x8 ; CHECK-OUTLINE-O0-NEXT: str w8, [sp, #12] -; CHECK-OUTLINE-O0-NEXT: mov w8, #4 ; =0x4 -; CHECK-OUTLINE-O0-NEXT: mov w0, w8 -; CHECK-OUTLINE-O0-NEXT: mov w3, #5 ; =0x5 ; CHECK-OUTLINE-O0-NEXT: bl ___atomic_store ; CHECK-OUTLINE-O0-NEXT: ldp x29, x30, [sp, #16] ; 16-byte Folded Reload ; CHECK-OUTLINE-O0-NEXT: add sp, sp, #32 @@ -2139,14 +2147,17 @@ define void @store_trunc(i32 %val, ptr %p8, ptr %p16, i64 %inp64) { ; CHECK-LSE-O0-NEXT: .cfi_def_cfa_offset 32 ; CHECK-LSE-O0-NEXT: .cfi_offset w30, -8 ; CHECK-LSE-O0-NEXT: .cfi_offset w29, -16 -; CHECK-LSE-O0-NEXT: stlrb w0, [x1] -; CHECK-LSE-O0-NEXT: strh w0, [x2] -; CHECK-LSE-O0-NEXT: add x2, sp, #12 +; CHECK-LSE-O0-NEXT: mov x9, x0 +; CHECK-LSE-O0-NEXT: mov x10, x2 ; CHECK-LSE-O0-NEXT: mov x8, x3 -; CHECK-LSE-O0-NEXT: str w8, [sp, #12] -; CHECK-LSE-O0-NEXT: mov w8, #4 ; =0x4 -; CHECK-LSE-O0-NEXT: mov x0, x8 +; CHECK-LSE-O0-NEXT: mov w11, #4 ; =0x4 +; CHECK-LSE-O0-NEXT: mov x0, x11 ; CHECK-LSE-O0-NEXT: mov w3, #5 ; =0x5 +; CHECK-LSE-O0-NEXT: add x2, sp, #12 +; CHECK-LSE-O0-NEXT: stlrb w9, [x1] +; CHECK-LSE-O0-NEXT: strh w9, [x10] +; CHECK-LSE-O0-NEXT: ; kill: def $w8 killed $w8 killed $x8 +; CHECK-LSE-O0-NEXT: str w8, [sp, #12] ; CHECK-LSE-O0-NEXT: bl ___atomic_store ; CHECK-LSE-O0-NEXT: ldp x29, x30, [sp, #16] ; 16-byte Folded Reload ; CHECK-LSE-O0-NEXT: add sp, sp, #32 diff --git a/llvm/test/CodeGen/AArch64/GlobalISel/atomic-anyextending-load-crash.ll b/llvm/test/CodeGen/AArch64/GlobalISel/atomic-anyextending-load-crash.ll index a3d57f05d9c60..2fa022d06da1f 100644 --- a/llvm/test/CodeGen/AArch64/GlobalISel/atomic-anyextending-load-crash.ll +++ b/llvm/test/CodeGen/AArch64/GlobalISel/atomic-anyextending-load-crash.ll @@ -11,10 +11,10 @@ define void @test(ptr %0) { ; CHECK-NEXT: .cfi_def_cfa_offset 144 ; CHECK-NEXT: .cfi_offset w30, -8 ; CHECK-NEXT: .cfi_offset w29, -16 -; CHECK-NEXT: ldar w8, [x0] -; CHECK-NEXT: str w8, [sp, #116] ; 4-byte Spill ; CHECK-NEXT: mov x8, #0 ; =0x0 ; CHECK-NEXT: str x8, [sp, #120] ; 8-byte Spill +; CHECK-NEXT: ldar w9, [x0] +; CHECK-NEXT: str w9, [sp, #116] ; 4-byte Spill ; CHECK-NEXT: blr x8 ; CHECK-NEXT: ldr w11, [sp, #116] ; 4-byte Reload ; CHECK-NEXT: ldr x8, [sp, #120] ; 8-byte Reload diff --git a/llvm/test/CodeGen/AArch64/GlobalISel/invoke-region.ll b/llvm/test/CodeGen/AArch64/GlobalISel/invoke-region.ll index 6efad350b4a96..9b404114f9913 100644 --- a/llvm/test/CodeGen/AArch64/GlobalISel/invoke-region.ll +++ b/llvm/test/CodeGen/AArch64/GlobalISel/invoke-region.ll @@ -12,10 +12,10 @@ define i1 @test_lpad_phi_widen_into_pred() personality ptr @__gxx_personality_v0 ; CHECK: bb.1 (%ir-block.0): ; CHECK-NEXT: successors: %bb.3(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: {{ $}} - ; CHECK-NEXT: [[GV:%[0-9]+]]:_(p0) = G_GLOBAL_VALUE @global_var ; CHECK-NEXT: [[C:%[0-9]+]]:_(i32) = G_CONSTANT i32 42 - ; CHECK-NEXT: G_STORE [[C]](i32), [[GV]](p0) :: (store (i32) into @global_var) + ; CHECK-NEXT: [[GV:%[0-9]+]]:_(p0) = G_GLOBAL_VALUE @global_var ; CHECK-NEXT: [[C1:%[0-9]+]]:_(i32) = G_CONSTANT i32 11 + ; CHECK-NEXT: G_STORE [[C]](i32), [[GV]](p0) :: (store (i32) into @global_var) ; CHECK-NEXT: [[C2:%[0-9]+]]:_(i16) = G_CONSTANT i16 1 ; CHECK-NEXT: G_INVOKE_REGION_START ; CHECK-NEXT: EH_LABEL <mcsymbol > @@ -31,8 +31,7 @@ define i1 @test_lpad_phi_widen_into_pred() personality ptr @__gxx_personality_v0 ; CHECK-NEXT: {{ $}} ; CHECK-NEXT: [[PHI:%[0-9]+]]:_(i32) = G_PHI [[C1]](i32), %bb.1 ; CHECK-NEXT: EH_LABEL <mcsymbol > - ; CHECK-NEXT: [[GV1:%[0-9]+]]:_(p0) = G_GLOBAL_VALUE @global_var - ; CHECK-NEXT: G_STORE [[PHI]](i32), [[GV1]](p0) :: (store (i32) into @global_var) + ; CHECK-NEXT: G_STORE [[PHI]](i32), [[GV]](p0) :: (store (i32) into @global_var) ; CHECK-NEXT: [[C3:%[0-9]+]]:_(i16) = G_CONSTANT i16 0 ; CHECK-NEXT: G_BR %bb.3 ; CHECK-NEXT: {{ $}} @@ -67,12 +66,12 @@ define i1 @test_lpad_phi_widen_into_pred_ext(ptr %ptr) personality ptr @__gxx_pe ; CHECK-NEXT: liveins: $x0 ; CHECK-NEXT: {{ $}} ; CHECK-NEXT: [[COPY:%[0-9]+]]:_(p0) = COPY $x0 - ; CHECK-NEXT: [[GV:%[0-9]+]]:_(p0) = G_GLOBAL_VALUE @global_var ; CHECK-NEXT: [[C:%[0-9]+]]:_(i32) = G_CONSTANT i32 42 + ; CHECK-NEXT: [[GV:%[0-9]+]]:_(p0) = G_GLOBAL_VALUE @global_var + ; CHECK-NEXT: [[C1:%[0-9]+]]:_(i32) = G_CONSTANT i32 11 ; CHECK-NEXT: G_STORE [[C]](i32), [[GV]](p0) :: (store (i32) into @global_var) ; CHECK-NEXT: [[LOAD:%[0-9]+]]:_(i8) = G_LOAD [[COPY]](p0) :: (load (i8) from %ir.ptr) ; CHECK-NEXT: [[ASSERT_ZEXT:%[0-9]+]]:_(i8) = G_ASSERT_ZEXT [[LOAD]], 1 - ; CHECK-NEXT: [[C1:%[0-9]+]]:_(i32) = G_CONSTANT i32 11 ; CHECK-NEXT: [[ANYEXT:%[0-9]+]]:_(i16) = G_ANYEXT [[ASSERT_ZEXT]](i8) ; CHECK-NEXT: G_INVOKE_REGION_START ; CHECK-NEXT: EH_LABEL <mcsymbol > @@ -88,8 +87,7 @@ define i1 @test_lpad_phi_widen_into_pred_ext(ptr %ptr) personality ptr @__gxx_pe ; CHECK-NEXT: {{ $}} ; CHECK-NEXT: [[PHI:%[0-9]+]]:_(i32) = G_PHI [[C1]](i32), %bb.1 ; CHECK-NEXT: EH_LABEL <mcsymbol > - ; CHECK-NEXT: [[GV1:%[0-9]+]]:_(p0) = G_GLOBAL_VALUE @global_var - ; CHECK-NEXT: G_STORE [[PHI]](i32), [[GV1]](p0) :: (store (i32) into @global_var) + ; CHECK-NEXT: G_STORE [[PHI]](i32), [[GV]](p0) :: (store (i32) into @global_var) ; CHECK-NEXT: [[C2:%[0-9]+]]:_(i16) = G_CONSTANT i16 0 ; CHECK-NEXT: G_BR %bb.3 ; CHECK-NEXT: {{ $}} diff --git a/llvm/test/CodeGen/AArch64/GlobalISel/legalize-exceptions.ll b/llvm/test/CodeGen/AArch64/GlobalISel/legalize-exceptions.ll index cedc70d4a5357..eb77d7082ca9b 100644 --- a/llvm/test/CodeGen/AArch64/GlobalISel/legalize-exceptions.ll +++ b/llvm/test/CodeGen/AArch64/GlobalISel/legalize-exceptions.ll @@ -13,10 +13,12 @@ define void @bar() personality ptr @__gxx_personality_v0 { ; CHECK: bb.1 (%ir-block.0): ; CHECK-NEXT: successors: %bb.3(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: {{ $}} + ; CHECK-NEXT: [[C:%[0-9]+]]:_(i32) = G_CONSTANT i32 42 + ; CHECK-NEXT: [[FRAME_INDEX:%[0-9]+]]:_(p0) = G_FRAME_INDEX %stack.0.exn.slot + ; CHECK-NEXT: [[FRAME_INDEX1:%[0-9]+]]:_(p0) = G_FRAME_INDEX %stack.1.ehselector.slot ; CHECK-NEXT: G_INVOKE_REGION_START ; CHECK-NEXT: EH_LABEL <mcsymbol > ; CHECK-NEXT: ADJCALLSTACKDOWN 0, 0, implicit-def $sp, implicit $sp - ; CHECK-NEXT: [[C:%[0-9]+]]:_(i32) = G_CONSTANT i32 42 ; CHECK-NEXT: $w0 = COPY [[C]](i32) ; CHECK-NEXT: BL @foo, csr_darwin_aarch64_aapcs, implicit-def dead $lr, implicit $sp, implicit $w0, implicit-def $w0 ; CHECK-NEXT: ADJCALLSTACKUP 0, 0, implicit-def $sp, implicit $sp @@ -32,9 +34,7 @@ define void @bar() personality ptr @__gxx_personality_v0 { ; CHECK-NEXT: [[COPY1:%[0-9]+]]:_(p0) = COPY $x1 ; CHECK-NEXT: [[PTRTOINT:%[0-9]+]]:_(i64) = G_PTRTOINT [[COPY1]](p0) ; CHECK-NEXT: [[TRUNC:%[0-9]+]]:_(i32) = G_TRUNC [[PTRTOINT]](i64) - ; CHECK-NEXT: [[FRAME_INDEX:%[0-9]+]]:_(p0) = G_FRAME_INDEX %stack.0.exn.slot ; CHECK-NEXT: G_STORE [[COPY]](p0), [[FRAME_INDEX]](p0) :: (store (p0) into %ir.exn.slot) - ; CHECK-NEXT: [[FRAME_INDEX1:%[0-9]+]]:_(p0) = G_FRAME_INDEX %stack.1.ehselector.slot ; CHECK-NEXT: G_STORE [[TRUNC]](i32), [[FRAME_INDEX1]](p0) :: (store (i32) into %ir.ehselector.slot) ; CHECK-NEXT: G_BR %bb.4 ; CHECK-NEXT: {{ $}} @@ -42,8 +42,7 @@ define void @bar() personality ptr @__gxx_personality_v0 { ; CHECK-NEXT: RET_ReallyLR ; CHECK-NEXT: {{ $}} ; CHECK-NEXT: bb.4.eh.resume: - ; CHECK-NEXT: [[FRAME_INDEX2:%[0-9]+]]:_(p0) = G_FRAME_INDEX %stack.0.exn.slot - ; CHECK-NEXT: [[LOAD:%[0-9]+]]:_(p0) = G_LOAD [[FRAME_INDEX2]](p0) :: (load (p0) from %ir.exn.slot) + ; CHECK-NEXT: [[LOAD:%[0-9]+]]:_(p0) = G_LOAD [[FRAME_INDEX]](p0) :: (load (p0) from %ir.exn.slot) ; CHECK-NEXT: ADJCALLSTACKDOWN 0, 0, implicit-def $sp, implicit $sp ; CHECK-NEXT: $x0 = COPY [[LOAD]](p0) ; CHECK-NEXT: BL @_Unwind_Resume, csr_darwin_aarch64_aapcs, implicit-def dead $lr, implicit $sp, implicit $x0 diff --git a/llvm/test/CodeGen/AArch64/GlobalISel/localizer-arm64-tti.ll b/llvm/test/CodeGen/AArch64/GlobalISel/localizer-arm64-tti.ll index 4fa647223e9d0..2902d1dc73332 100644 --- a/llvm/test/CodeGen/AArch64/GlobalISel/localizer-arm64-tti.ll +++ b/llvm/test/CodeGen/AArch64/GlobalISel/localizer-arm64-tti.ll @@ -1,5 +1,5 @@ ; NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py -; RUN: llc -o - -verify-machineinstrs -O0 -global-isel -stop-after=localizer %s | FileCheck %s +; RUN: llc -o - -verify-machineinstrs -O1 -global-isel -stop-after=localizer %s | FileCheck %s target datalayout = "e-m:o-i64:64-i128:128-n32:64-S128" target triple = "arm64-apple-ios5.0.0" @@ -25,11 +25,11 @@ define i32 @foo() { ; CHECK-NEXT: [[GV1:%[0-9]+]]:_(p0) = G_GLOBAL_VALUE @var3 ; CHECK-NEXT: [[C2:%[0-9]+]]:_(i32) = G_CONSTANT i32 0 ; CHECK-NEXT: [[GV2:%[0-9]+]]:_(p0) = G_GLOBAL_VALUE @var1 - ; CHECK-NEXT: [[LOAD:%[0-9]+]]:_(i32) = G_LOAD [[GV2]](p0) :: (load (i32) from @var1) + ; CHECK-NEXT: [[LOAD:%[0-9]+]]:_(i32) = G_LOAD [[GV2]](p0) :: (dereferenceable load (i32) from @var1) ; CHECK-NEXT: [[C3:%[0-9]+]]:_(i32) = G_CONSTANT i32 1 - ; CHECK-NEXT: [[ICMP:%[0-9]+]]:_(i1) = G_ICMP intpred(eq), [[LOAD]](i32), [[C3]] - ; CHECK-NEXT: G_BRCOND [[ICMP]](i1), %bb.2 - ; CHECK-NEXT: G_BR %bb.3 + ; CHECK-NEXT: [[ICMP:%[0-9]+]]:_(i1) = G_ICMP intpred(ne), [[LOAD]](i32), [[C3]] + ; CHECK-NEXT: G_BRCOND [[ICMP]](i1), %bb.3 + ; CHECK-NEXT: G_BR %bb.2 ; CHECK-NEXT: {{ $}} ; CHECK-NEXT: bb.2.if.then: ; CHECK-NEXT: successors: %bb.3(0x80000000) @@ -43,7 +43,6 @@ define i32 @foo() { ; CHECK-NEXT: [[GV5:%[0-9]+]]:_(p0) = G_GLOBAL_VALUE @var3 ; CHECK-NEXT: G_STORE [[C4]](i32), [[GV5]](p0) :: (store (i32) into @var3) ; CHECK-NEXT: G_STORE [[C5]](i32), [[GV4]](p0) :: (store (i32) into @var1) - ; CHECK-NEXT: G_BR %bb.3 ; CHECK-NEXT: {{ $}} ; CHECK-NEXT: bb.3.if.end: ; CHECK-NEXT: [[C6:%[0-9]+]]:_(i32) = G_CONSTANT i32 0 @@ -68,35 +67,29 @@ if.end: define i32 @imm_cost_too_large_cost_of_2() { ; CHECK-LABEL: name: imm_cost_too_large_cost_of_2 ; CHECK: bb.1.entry: - ; CHECK-NEXT: successors: %bb.2(0x40000000), %bb.4(0x40000000) + ; CHECK-NEXT: successors: %bb.2(0x40000000), %bb.3(0x40000000) ; CHECK-NEXT: {{ $}} ; CHECK-NEXT: [[GV:%[0-9]+]]:_(p0) = G_GLOBAL_VALUE @var2 ; CHECK-NEXT: [[GV1:%[0-9]+]]:_(p0) = G_GLOBAL_VALUE @var3 ; CHECK-NEXT: [[C:%[0-9]+]]:_(i32) = G_CONSTANT i32 0 - ; CHECK-NEXT: [[GV2:%[0-9]+]]:_(p0) = G_GLOBAL_VALUE @var1 - ; CHECK-NEXT: [[LOAD:%[0-9]+]]:_(i32) = G_LOAD [[GV2]](p0) :: (load (i32) from @var1) ; CHECK-NEXT: [[C1:%[0-9]+]]:_(i32) = G_CONSTANT i32 -2228259 ; CHECK-NEXT: [[CONSTANT_FOLD_BARRIER:%[0-9]+]]:_(i32) = G_CONSTANT_FOLD_BARRIER [[C1]] + ; CHECK-NEXT: [[GV2:%[0-9]+]]:_(p0) = G_GLOBAL_VALUE @var1 + ; CHECK-NEXT: [[LOAD:%[0-9]+]]:_(i32) = G_LOAD [[GV2]](p0) :: (dereferenceable load (i32) from @var1) ; CHECK-NEXT: [[C2:%[0-9]+]]:_(i32) = G_CONSTANT i32 1 - ; CHECK-NEXT: [[ICMP:%[0-9]+]]:_(i1) = G_ICMP intpred(eq), [[LOAD]](i32), [[C2]] - ; CHECK-NEXT: G_BRCOND [[ICMP]](i1), %bb.2 - ; CHECK-NEXT: G_BR %bb.4 + ; CHECK-NEXT: [[ICMP:%[0-9]+]]:_(i1) = G_ICMP intpred(ne), [[LOAD]](i32), [[C2]] + ; CHECK-NEXT: G_BRCOND [[ICMP]](i1), %bb.3 + ; CHECK-NEXT: G_BR %bb.2 ; CHECK-NEXT: {{ $}} ; CHECK-NEXT: bb.2.if.then: ; CHECK-NEXT: successors: %bb.3(0x80000000) ; CHECK-NEXT: {{ $}} ; CHECK-NEXT: [[GV3:%[0-9]+]]:_(p0) = G_GLOBAL_VALUE @var2 ; CHECK-NEXT: G_STORE [[CONSTANT_FOLD_BARRIER]](i32), [[GV3]](p0) :: (store (i32) into @var2) - ; CHECK-NEXT: G_BR %bb.3 - ; CHECK-NEXT: {{ $}} - ; CHECK-NEXT: bb.3.if.then2: - ; CHECK-NEXT: successors: %bb.4(0x80000000) - ; CHECK-NEXT: {{ $}} ; CHECK-NEXT: [[GV4:%[0-9]+]]:_(p0) = G_GLOBAL_VALUE @var1 ; CHECK-NEXT: G_STORE [[CONSTANT_FOLD_BARRIER]](i32), [[GV4]](p0) :: (store (i32) into @var1) - ; CHECK-NEXT: G_BR %bb.4 ; CHECK-NEXT: {{ $}} - ; CHECK-NEXT: bb.4.if.end: + ; CHECK-NEXT: bb.3.if.end: ; CHECK-NEXT: [[GV5:%[0-9]+]]:_(p0) = G_GLOBAL_VALUE @var3 ; CHECK-NEXT: G_STORE [[CONSTANT_FOLD_BARRIER]](i32), [[GV5]](p0) :: (store (i32) into @var3) ; CHECK-NEXT: [[C3:%[0-9]+]]:_(i32) = G_CONSTANT i32 0 @@ -124,35 +117,29 @@ if.end: define i64 @imm_cost_too_large_cost_of_4() { ; CHECK-LABEL: name: imm_cost_too_large_cost_of_4 ; CHECK: bb.1.entry: - ; CHECK-NEXT: successors: %bb.2(0x40000000), %bb.4(0x40000000) + ; CHECK-NEXT: successors: %bb.2(0x40000000), %bb.3(0x40000000) ; CHECK-NEXT: {{ $}} ; CHECK-NEXT: [[GV:%[0-9]+]]:_(p0) = G_GLOBAL_VALUE @var2_64 ; CHECK-NEXT: [[GV1:%[0-9]+]]:_(p0) = G_GLOBAL_VALUE @var3_64 ; CHECK-NEXT: [[C:%[0-9]+]]:_(i64) = G_CONSTANT i64 0 - ; CHECK-NEXT: [[GV2:%[0-9]+]]:_(p0) = G_GLOBAL_VALUE @var1_64 - ; CHECK-NEXT: [[LOAD:%[0-9]+]]:_(i64) = G_LOAD [[GV2]](p0) :: (load (i64) from @var1_64, align 4) ; CHECK-NEXT: [[C1:%[0-9]+]]:_(i64) = G_CONSTANT i64 -2228259 ; CHECK-NEXT: [[CONSTANT_FOLD_BARRIER:%[0-9]+]]:_(i64) = G_CONSTANT_FOLD_BARRIER [[C1]] + ; CHECK-NEXT: [[GV2:%[0-9]+]]:_(p0) = G_GLOBAL_VALUE @var1_64 + ; CHECK-NEXT: [[LOAD:%[0-9]+]]:_(i64) = G_LOAD [[GV2]](p0) :: (dereferenceable load (i64) from @var1_64, align 4) ; CHECK-NEXT: [[C2:%[0-9]+]]:_(i64) = G_CONSTANT i64 1 - ; CHECK-NEXT: [[ICMP:%[0-9]+]]:_(i1) = G_ICMP intpred(eq), [[LOAD]](i64), [[C2]] - ; CHECK-NEXT: G_BRCOND [[ICMP]](i1), %bb.2 - ; CHECK-NEXT: G_BR %bb.4 + ; CHECK-NEXT: [[ICMP:%[0-9]+]]:_(i1) = G_ICMP intpred(ne), [[LOAD]](i64), [[C2]] + ; CHECK-NEXT: G_BRCOND [[ICMP]](i1), %bb.3 + ; CHECK-NEXT: G_BR %bb.2 ; CHECK-NEXT: {{ $}} ; CHECK-NEXT: bb.2.if.then: ; CHECK-NEXT: successors: %bb.3(0x80000000) ; CHECK-NEXT: {{ $}} ; CHECK-NEXT: [[GV3:%[0-9]+]]:_(p0) = G_GLOBAL_VALUE @var2_64 ; CHECK-NEXT: G_STORE [[CONSTANT_FOLD_BARRIER]](i64), [[GV3]](p0) :: (store (i64) into @var2_64) - ; CHECK-NEXT: G_BR %bb.3 - ; CHECK-NEXT: {{ $}} - ; CHECK-NEXT: bb.3.if.then2: - ; CHECK-NEXT: successors: %bb.4(0x80000000) - ; CHECK-NEXT: {{ $}} ; CHECK-NEXT: [[GV4:%[0-9]+]]:_(p0) = G_GLOBAL_VALUE @var1_64 ; CHECK-NEXT: G_STORE [[CONSTANT_FOLD_BARRIER]](i64), [[GV4]](p0) :: (store (i64) into @var1_64) - ; CHECK-NEXT: G_BR %bb.4 ; CHECK-NEXT: {{ $}} - ; CHECK-NEXT: bb.4.if.end: + ; CHECK-NEXT: bb.3.if.end: ; CHECK-NEXT: [[GV5:%[0-9]+]]:_(p0) = G_GLOBAL_VALUE @var3_64 ; CHECK-NEXT: G_STORE [[CONSTANT_FOLD_BARRIER]](i64), [[GV5]](p0) :: (store (i64) into @var3_64) ; CHECK-NEXT: [[C3:%[0-9]+]]:_(i64) = G_CONSTANT i64 0 @@ -180,7 +167,7 @@ if.end: define i64 @f64_imm_cost_too_high(double %a) { ; CHECK-LABEL: name: f64_imm_cost_too_high ; CHECK: bb.1.entry: - ; CHECK-NEXT: successors: %bb.2(0x40000000), %bb.4(0x40000000) + ; CHECK-NEXT: successors: %bb.2(0x40000000), %bb.3(0x40000000) ; CHECK-NEXT: liveins: $d0 ; CHECK-NEXT: {{ $}} ; CHECK-NEXT: [[C:%[0-9]+]]:_(f64) = G_FCONSTANT double 1.000000e-02 @@ -188,27 +175,21 @@ define i64 @f64_imm_cost_too_high(double %a) { ; CHECK-NEXT: [[GV1:%[0-9]+]]:_(p0) = G_GLOBAL_VALUE @var3_64 ; CHECK-NEXT: [[C1:%[0-9]+]]:_(i64) = G_CONSTANT i64 0 ; CHECK-NEXT: [[GV2:%[0-9]+]]:_(p0) = G_GLOBAL_VALUE @var1_64 - ; CHECK-NEXT: [[LOAD:%[0-9]+]]:_(i64) = G_LOAD [[GV2]](p0) :: (load (i64) from @var1_64, align 4) + ; CHECK-NEXT: [[LOAD:%[0-9]+]]:_(i64) = G_LOAD [[GV2]](p0) :: (dereferenceable load (i64) from @var1_64, align 4) ; CHECK-NEXT: [[C2:%[0-9]+]]:_(i64) = G_CONSTANT i64 1 - ; CHECK-NEXT: [[ICMP:%[0-9]+]]:_(i1) = G_ICMP intpred(eq), [[LOAD]](i64), [[C2]] - ; CHECK-NEXT: G_BRCOND [[ICMP]](i1), %bb.2 - ; CHECK-NEXT: G_BR %bb.4 + ; CHECK-NEXT: [[ICMP:%[0-9]+]]:_(i1) = G_ICMP intpred(ne), [[LOAD]](i64), [[C2]] + ; CHECK-NEXT: G_BRCOND [[ICMP]](i1), %bb.3 + ; CHECK-NEXT: G_BR %bb.2 ; CHECK-NEXT: {{ $}} ; CHECK-NEXT: bb.2.if.then: ; CHECK-NEXT: successors: %bb.3(0x80000000) ; CHECK-NEXT: {{ $}} ; CHECK-NEXT: [[GV3:%[0-9]+]]:_(p0) = G_GLOBAL_VALUE @var2_64 ; CHECK-NEXT: G_STORE [[C]](f64), [[GV3]](p0) :: (store (f64) into @var2_64) - ; CHECK-NEXT: G_BR %bb.3 - ; CHECK-NEXT: {{ $}} - ; CHECK-NEXT: bb.3.if.then2: - ; CHECK-NEXT: successors: %bb.4(0x80000000) - ; CHECK-NEXT: {{ $}} ; CHECK-NEXT: [[GV4:%[0-9]+]]:_(p0) = G_GLOBAL_VALUE @var1_64 ; CHECK-NEXT: G_STORE [[C]](f64), [[GV4]](p0) :: (store (f64) into @var1_64) - ; CHECK-NEXT: G_BR %bb.4 ; CHECK-NEXT: {{ $}} - ; CHECK-NEXT: bb.4.if.end: + ; CHECK-NEXT: bb.3.if.end: ; CHECK-NEXT: [[GV5:%[0-9]+]]:_(p0) = G_GLOBAL_VALUE @var3_64 ; CHECK-NEXT: G_STORE [[C]](f64), [[GV5]](p0) :: (store (f64) into @var3_64) ; CHECK-NEXT: [[C3:%[0-9]+]]:_(i64) = G_CONSTANT i64 0 @@ -235,7 +216,7 @@ if.end: define i64 @f64_imm_cheap(double %a) { ; CHECK-LABEL: name: f64_imm_cheap ; CHECK: bb.1.entry: - ; CHECK-NEXT: successors: %bb.2(0x40000000), %bb.4(0x40000000) + ; CHECK-NEXT: successors: %bb.2(0x40000000), %bb.3(0x40000000) ; CHECK-NEXT: liveins: $d0 ; CHECK-NEXT: {{ $}} ; CHECK-NEXT: [[C:%[0-9]+]]:_(f64) = G_FCONSTANT double 0.000000e+00 @@ -243,11 +224,11 @@ define i64 @f64_imm_cheap(double %a) { ; CHECK-NEXT: [[GV1:%[0-9]+]]:_(p0) = G_GLOBAL_VALUE @var3_64 ; CHECK-NEXT: [[C1:%[0-9]+]]:_(i64) = G_CONSTANT i64 0 ; CHECK-NEXT: [[GV2:%[0-9]+]]:_(p0) = G_GLOBAL_VALUE @var1_64 - ; CHECK-NEXT: [[LOAD:%[0-9]+]]:_(i64) = G_LOAD [[GV2]](p0) :: (load (i64) from @var1_64, align 4) + ; CHECK-NEXT: [[LOAD:%[0-9]+]]:_(i64) = G_LOAD [[GV2]](p0) :: (dereferenceable load (i64) from @var1_64, align 4) ; CHECK-NEXT: [[C2:%[0-9]+]]:_(i64) = G_CONSTANT i64 1 - ; CHECK-NEXT: [[ICMP:%[0-9]+]]:_(i1) = G_ICMP intpred(eq), [[LOAD]](i64), [[C2]] - ; CHECK-NEXT: G_BRCOND [[ICMP]](i1), %bb.2 - ; CHECK-NEXT: G_BR %bb.4 + ; CHECK-NEXT: [[ICMP:%[0-9]+]]:_(i1) = G_ICMP intpred(ne), [[LOAD]](i64), [[C2]] + ; CHECK-NEXT: G_BRCOND [[ICMP]](i1), %bb.3 + ; CHECK-NEXT: G_BR %bb.2 ; CHECK-NEXT: {{ $}} ; CHECK-NEXT: bb.2.if.then: ; CHECK-NEXT: successors: %bb.3(0x80000000) @@ -255,22 +236,15 @@ define i64 @f64_imm_cheap(double %a) { ; CHECK-NEXT: [[GV3:%[0-9]+]]:_(p0) = G_GLOBAL_VALUE @var2_64 ; CHECK-NEXT: [[C3:%[0-9]+]]:_(f64) = G_FCONSTANT double 0.000000e+00 ; CHECK-NEXT: G_STORE [[C3]](f64), [[GV3]](p0) :: (store (f64) into @var2_64) - ; CHECK-NEXT: G_BR %bb.3 - ; CHECK-NEXT: {{ $}} - ; CHECK-NEXT: bb.3.if.then2: - ; CHECK-NEXT: successors: %bb.4(0x80000000) - ; CHECK-NEXT: {{ $}} - ; CHECK-NEXT: [[C4:%[0-9]+]]:_(f64) = G_FCONSTANT double 0.000000e+00 ; CHECK-NEXT: [[GV4:%[0-9]+]]:_(p0) = G_GLOBAL_VALUE @var1_64 - ; CHECK-NEXT: G_STORE [[C4]](f64), [[GV4]](p0) :: (store (f64) into @var1_64) - ; CHECK-NEXT: G_BR %bb.4 + ; CHECK-NEXT: G_STORE [[C3]](f64), [[GV4]](p0) :: (store (f64) into @var1_64) ; CHECK-NEXT: {{ $}} - ; CHECK-NEXT: bb.4.if.end: + ; CHECK-NEXT: bb.3.if.end: ; CHECK-NEXT: [[GV5:%[0-9]+]]:_(p0) = G_GLOBAL_VALUE @var3_64 - ; CHECK-NEXT: [[C5:%[0-9]+]]:_(f64) = G_FCONSTANT double 0.000000e+00 - ; CHECK-NEXT: G_STORE [[C5]](f64), [[GV5]](p0) :: (store (f64) into @var3_64) - ; CHECK-NEXT: [[C6:%[0-9]+]]:_(i64) = G_CONSTANT i64 0 - ; CHECK-NEXT: $x0 = COPY [[C6]](i64) + ; CHECK-NEXT: [[C4:%[0-9]+]]:_(f64) = G_FCONSTANT double 0.000000e+00 + ; CHECK-NEXT: G_STORE [[C4]](f64), [[GV5]](p0) :: (store (f64) into @var3_64) + ; CHECK-NEXT: [[C5:%[0-9]+]]:_(i64) = G_CONSTANT i64 0 + ; CHECK-NEXT: $x0 = COPY [[C5]](i64) ; CHECK-NEXT: RET_ReallyLR implicit $x0 entry: %0 = load i64, ptr @var1_64, align 4 diff --git a/llvm/test/CodeGen/AArch64/GlobalISel/localizer-propagate-debug-loc.mir b/llvm/test/CodeGen/AArch64/GlobalISel/localizer-propagate-debug-loc.mir index 6ced34daed680..e20aebb84be1d 100644 --- a/llvm/test/CodeGen/AArch64/GlobalISel/localizer-propagate-debug-loc.mir +++ b/llvm/test/CodeGen/AArch64/GlobalISel/localizer-propagate-debug-loc.mir @@ -1,6 +1,5 @@ # NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py -# RUN: llc -O0 %s -global-isel -start-before localizer \ -# RUN: -stop-after localizer -o - | FileCheck --check-prefix=CHECK %s +# RUN: llc -run-pass=localizer %s -o - | FileCheck --check-prefix=CHECK %s --- | target datalayout = "e-m:o-i64:64-i128:128-n32:64-S128" target triple = "arm64-apple-macosx12.0.0" @@ -76,12 +75,12 @@ body: | ; CHECK-NEXT: successors: %bb.1(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: {{ $}} ; CHECK-NEXT: [[C:%[0-9]+]]:gpr(i32) = G_CONSTANT i32 3 - ; CHECK-NEXT: [[ADRP:%[0-9]+]]:gpr64(p0) = ADRP target-flags(aarch64-page) @C, debug-location !DILocation(line: 0, scope: !18) - ; CHECK-NEXT: [[ADD_LOW:%[0-9]+]]:gpr(p0) = G_ADD_LOW [[ADRP]](p0), target-flags(aarch64-pageoff, aarch64-nc) @C, debug-location !DILocation(line: 0, scope: !18) - ; CHECK-NEXT: [[ADRP1:%[0-9]+]]:gpr64(p0) = ADRP target-flags(aarch64-page) @B, debug-location !DILocation(line: 0, scope: !14) - ; CHECK-NEXT: [[ADD_LOW1:%[0-9]+]]:gpr(p0) = G_ADD_LOW [[ADRP1]](p0), target-flags(aarch64-pageoff, aarch64-nc) @B, debug-location !DILocation(line: 0, scope: !14) - ; CHECK-NEXT: [[ADRP2:%[0-9]+]]:gpr64(p0) = ADRP target-flags(aarch64-page) @A, debug-location !DILocation(line: 0, scope: !11) - ; CHECK-NEXT: [[ADD_LOW2:%[0-9]+]]:gpr(p0) = G_ADD_LOW [[ADRP2]](p0), target-flags(aarch64-pageoff, aarch64-nc) @A, debug-location !DILocation(line: 0, scope: !11) + ; CHECK-NEXT: [[ADRP:%[0-9]+]]:gpr64(p0) = ADRP target-flags(aarch64-page) @C, debug-location !DILocation(line: 0, scope: !18) + ; CHECK-NEXT: [[ADD_LOW:%[0-9]+]]:gpr(p0) = G_ADD_LOW [[ADRP]](p0), target-flags(aarch64-pageoff, aarch64-nc) @C, debug-location !DILocation(line: 0, scope: !18) + ; CHECK-NEXT: [[ADRP1:%[0-9]+]]:gpr64(p0) = ADRP target-flags(aarch64-page) @B, debug-location !DILocation(line: 0, scope: !14) + ; CHECK-NEXT: [[ADD_LOW1:%[0-9]+]]:gpr(p0) = G_ADD_LOW [[ADRP1]](p0), target-flags(aarch64-pageoff, aarch64-nc) @B, debug-location !DILocation(line: 0, scope: !14) + ; CHECK-NEXT: [[ADRP2:%[0-9]+]]:gpr64(p0) = ADRP target-flags(aarch64-page) @A, debug-location !DILocation(line: 0, scope: !11) + ; CHECK-NEXT: [[ADD_LOW2:%[0-9]+]]:gpr(p0) = G_ADD_LOW [[ADRP2]](p0), target-flags(aarch64-pageoff, aarch64-nc) @A, debug-location !DILocation(line: 0, scope: !11) ; CHECK-NEXT: [[FRAME_INDEX:%[0-9]+]]:gpr(p0) = G_FRAME_INDEX %stack.0 ; CHECK-NEXT: [[C1:%[0-9]+]]:gpr(i32) = G_CONSTANT i32 0 ; CHECK-NEXT: [[COPY:%[0-9]+]]:gpr(i32) = COPY [[C1]](i32) @@ -94,9 +93,9 @@ body: | ; CHECK-NEXT: bb.1: ; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: {{ $}} - ; CHECK-NEXT: [[ADRP3:%[0-9]+]]:gpr64(p0) = ADRP target-flags(aarch64-page) @A, debug-location !10 - ; CHECK-NEXT: [[ADD_LOW3:%[0-9]+]]:gpr(p0) = G_ADD_LOW [[ADRP3]](p0), target-flags(aarch64-pageoff, aarch64-nc) @A, debug-location !10 - ; CHECK-NEXT: [[LOAD:%[0-9]+]]:gpr(i32) = G_LOAD [[ADD_LOW3]](p0), debug-location !10 :: (dereferenceable load (i32)) + ; CHECK-NEXT: [[ADRP3:%[0-9]+]]:gpr64(p0) = ADRP target-flags(aarch64-page) @A, debug-location !10 + ; CHECK-NEXT: [[ADD_LOW3:%[0-9]+]]:gpr(p0) = G_ADD_LOW [[ADRP3]](p0), target-flags(aarch64-pageoff, aarch64-nc) @A, debug-location !10 + ; CHECK-NEXT: [[LOAD:%[0-9]+]]:gpr(i32) = G_LOAD [[ADD_LOW3]](p0), debug-location !10 :: (dereferenceable load (i32)) ; CHECK-NEXT: [[FRAME_INDEX1:%[0-9]+]]:gpr(p0) = G_FRAME_INDEX %stack.0 ; CHECK-NEXT: G_STORE [[LOAD]](i32), [[FRAME_INDEX1]](p0) :: (volatile store (i32) into %ir.1) ; CHECK-NEXT: G_BR %bb.5 @@ -114,22 +113,22 @@ body: | ; CHECK-NEXT: bb.3: ; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: {{ $}} - ; CHECK-NEXT: [[ADRP4:%[0-9]+]]:gpr64(p0) = ADRP target-flags(aarch64-page) @B, debug-location !DILocation(line: 0, scope: !14) - ; CHECK-NEXT: [[ADD_LOW4:%[0-9]+]]:gpr(p0) = G_ADD_LOW [[ADRP4]](p0), target-flags(aarch64-pageoff, aarch64-nc) @B, debug-location !DILocation(line: 0, scope: !14) - ; CHECK-NEXT: [[LOAD1:%[0-9]+]]:gpr(i32) = G_LOAD [[ADD_LOW4]](p0), debug-location !13 :: (dereferenceable load (i32)) + ; CHECK-NEXT: [[ADRP4:%[0-9]+]]:gpr64(p0) = ADRP target-flags(aarch64-page) @B, debug-location !DILocation(line: 0, scope: !14) + ; CHECK-NEXT: [[ADD_LOW4:%[0-9]+]]:gpr(p0) = G_ADD_LOW [[ADRP4]](p0), target-flags(aarch64-pageoff, aarch64-nc) @B, debug-location !DILocation(line: 0, scope: !14) + ; CHECK-NEXT: [[LOAD1:%[0-9]+]]:gpr(i32) = G_LOAD [[ADD_LOW4]](p0), debug-location !13 :: (dereferenceable load (i32)) ; CHECK-NEXT: [[FRAME_INDEX2:%[0-9]+]]:gpr(p0) = G_FRAME_INDEX %stack.0 ; CHECK-NEXT: G_STORE [[LOAD1]](i32), [[FRAME_INDEX2]](p0) :: (volatile store (i32) into %ir.1) - ; CHECK-NEXT: [[LOAD2:%[0-9]+]]:gpr(i32) = G_LOAD [[ADD_LOW4]](p0), debug-location !16 :: (dereferenceable load (i32)) + ; CHECK-NEXT: [[LOAD2:%[0-9]+]]:gpr(i32) = G_LOAD [[ADD_LOW4]](p0), debug-location !16 :: (dereferenceable load (i32)) ; CHECK-NEXT: G_STORE [[LOAD2]](i32), [[FRAME_INDEX2]](p0) :: (volatile store (i32) into %ir.1) ; CHECK-NEXT: G_BR %bb.5 ; CHECK-NEXT: {{ $}} ; CHECK-NEXT: bb.4: ; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: {{ $}} - ; CHECK-NEXT: [[ADRP5:%[0-9]+]]:gpr64(p0) = ADRP target-flags(aarch64-page) @C, debug-location !17 - ; CHECK-NEXT: [[ADD_LOW5:%[0-9]+]]:gpr(p0) = G_ADD_LOW [[ADRP5]](p0), target-flags(aarch64-pageoff, aarch64-nc) @C, debug-location !17 - ; CHECK-NEXT: [[C5:%[0-9]+]]:gpr(i32) = G_CONSTANT i32 3, debug-location !17 - ; CHECK-NEXT: G_STORE [[C5]](i32), [[ADD_LOW5]](p0), debug-location !17 :: (store (i32) into @C) + ; CHECK-NEXT: [[ADRP5:%[0-9]+]]:gpr64(p0) = ADRP target-flags(aarch64-page) @C, debug-location !17 + ; CHECK-NEXT: [[ADD_LOW5:%[0-9]+]]:gpr(p0) = G_ADD_LOW [[ADRP5]](p0), target-flags(aarch64-pageoff, aarch64-nc) @C, debug-location !17 + ; CHECK-NEXT: [[C5:%[0-9]+]]:gpr(i32) = G_CONSTANT i32 3, debug-location !17 + ; CHECK-NEXT: G_STORE [[C5]](i32), [[ADD_LOW5]](p0), debug-location !17 :: (store (i32) into @C) ; CHECK-NEXT: G_BR %bb.5 ; CHECK-NEXT: {{ $}} ; CHECK-NEXT: bb.5: diff --git a/llvm/test/CodeGen/AArch64/GlobalISel/localizer-tls.ll b/llvm/test/CodeGen/AArch64/GlobalISel/localizer-tls.ll index b8f65310ae998..b2d2f4e17c327 100644 --- a/llvm/test/CodeGen/AArch64/GlobalISel/localizer-tls.ll +++ b/llvm/test/CodeGen/AArch64/GlobalISel/localizer-tls.ll @@ -1,6 +1,6 @@ ; NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py -; RUN: llc -mtriple=arm64-apple-ios -o - -verify-machineinstrs -O0 -global-isel -stop-after=localizer %s | FileCheck %s -; RUN: llc -mtriple=aarch64-linux-gnu -o - -verify-machineinstrs -O0 -global-isel -stop-after=localizer %s | FileCheck %s +; RUN: llc -mtriple=arm64-apple-ios -o - -verify-machineinstrs -O1 -global-isel -stop-after=localizer %s | FileCheck %s +; RUN: llc -mtriple=aarch64-linux-gnu -o - -verify-machineinstrs -O1 -global-isel -stop-after=localizer %s | FileCheck %s @var1 = common global i32 0, align 4 @var2 = common global i32 0, align 4 @@ -20,19 +20,18 @@ define i32 @tls() { ; CHECK-NEXT: [[GV1:%[0-9]+]]:_(p0) = G_GLOBAL_VALUE @var2 ; CHECK-NEXT: [[C:%[0-9]+]]:_(i32) = G_CONSTANT i32 0 ; CHECK-NEXT: [[GV2:%[0-9]+]]:_(p0) = G_GLOBAL_VALUE @var1 - ; CHECK-NEXT: [[LOAD:%[0-9]+]]:_(i32) = G_LOAD [[GV2]](p0) :: (load (i32) from @var1) + ; CHECK-NEXT: [[LOAD:%[0-9]+]]:_(i32) = G_LOAD [[GV2]](p0) :: (dereferenceable load (i32) from @var1) ; CHECK-NEXT: [[C1:%[0-9]+]]:_(i32) = G_CONSTANT i32 1 - ; CHECK-NEXT: [[ICMP:%[0-9]+]]:_(i1) = G_ICMP intpred(eq), [[LOAD]](i32), [[C1]] - ; CHECK-NEXT: G_BRCOND [[ICMP]](i1), %bb.2 - ; CHECK-NEXT: G_BR %bb.3 + ; CHECK-NEXT: [[ICMP:%[0-9]+]]:_(i1) = G_ICMP intpred(ne), [[LOAD]](i32), [[C1]] + ; CHECK-NEXT: G_BRCOND [[ICMP]](i1), %bb.3 + ; CHECK-NEXT: G_BR %bb.2 ; CHECK-NEXT: {{ $}} ; CHECK-NEXT: bb.2.if.then: ; CHECK-NEXT: successors: %bb.3(0x80000000) ; CHECK-NEXT: {{ $}} - ; CHECK-NEXT: [[LOAD1:%[0-9]+]]:_(i32) = G_LOAD [[GV]](p0) :: (load (i32) from @tls_gv) + ; CHECK-NEXT: [[LOAD1:%[0-9]+]]:_(i32) = G_LOAD [[GV]](p0) :: (dereferenceable load (i32) from @tls_gv) ; CHECK-NEXT: [[GV3:%[0-9]+]]:_(p0) = G_GLOBAL_VALUE @var2 ; CHECK-NEXT: G_STORE [[LOAD1]](i32), [[GV3]](p0) :: (store (i32) into @var2) - ; CHECK-NEXT: G_BR %bb.3 ; CHECK-NEXT: {{ $}} ; CHECK-NEXT: bb.3.if.end: ; CHECK-NEXT: [[C2:%[0-9]+]]:_(i32) = G_CONSTANT i32 0 @@ -51,4 +50,3 @@ if.then: if.end: ret i32 0 } - diff --git a/llvm/test/CodeGen/AArch64/GlobalISel/select-fp-anyext-crash.ll b/llvm/test/CodeGen/AArch64/GlobalISel/select-fp-anyext-crash.ll index a63636e666e8f..4b57ef1d2d24f 100644 --- a/llvm/test/CodeGen/AArch64/GlobalISel/select-fp-anyext-crash.ll +++ b/llvm/test/CodeGen/AArch64/GlobalISel/select-fp-anyext-crash.ll @@ -13,19 +13,19 @@ define i32 @test() { ; CHECK-NEXT: .cfi_def_cfa_offset 80 ; CHECK-NEXT: .cfi_offset w30, -8 ; CHECK-NEXT: .cfi_offset w29, -16 +; CHECK-NEXT: mov x9, #0 ; =0x0 ; CHECK-NEXT: mov x8, #0 ; =0x0 -; CHECK-NEXT: ldr s0, [x8] +; CHECK-NEXT: mov w10, #0 ; =0x0 +; CHECK-NEXT: str w10, [sp, #60] ; 4-byte Spill +; CHECK-NEXT: ldr s0, [x9] ; CHECK-NEXT: ; kill: def $d0 killed $s0 -; CHECK-NEXT: mov x8, sp -; CHECK-NEXT: mov w9, #0 ; =0x0 -; CHECK-NEXT: str w9, [sp, #60] ; 4-byte Spill -; CHECK-NEXT: str xzr, [x8] -; CHECK-NEXT: str xzr, [x8, #8] -; CHECK-NEXT: str xzr, [x8, #16] -; CHECK-NEXT: str xzr, [x8, #24] -; CHECK-NEXT: str d0, [x8, #32] -; CHECK-NEXT: str xzr, [x8, #40] -; CHECK-NEXT: mov x8, #0 ; =0x0 +; CHECK-NEXT: mov x9, sp +; CHECK-NEXT: str xzr, [x9] +; CHECK-NEXT: str xzr, [x9, #8] +; CHECK-NEXT: str xzr, [x9, #16] +; CHECK-NEXT: str xzr, [x9, #24] +; CHECK-NEXT: str d0, [x9, #32] +; CHECK-NEXT: str xzr, [x9, #40] ; CHECK-NEXT: mov x0, x8 ; CHECK-NEXT: blr x8 ; CHECK-NEXT: ldr w0, [sp, #60] ; 4-byte Reload diff --git a/llvm/test/CodeGen/AArch64/GlobalISel/translate-constant-dag.ll b/llvm/test/CodeGen/AArch64/GlobalISel/translate-constant-dag.ll index e7592fe49c863..11a768fcdbf47 100644 --- a/llvm/test/CodeGen/AArch64/GlobalISel/translate-constant-dag.ll +++ b/llvm/test/CodeGen/AArch64/GlobalISel/translate-constant-dag.ll @@ -10,12 +10,12 @@ define void @test_const(ptr %dst) { ; CHECK-NEXT: {{ $}} ; CHECK-NEXT: [[COPY:%[0-9]+]]:gpr64sp = COPY $x0 ; CHECK-NEXT: [[MOVi32imm:%[0-9]+]]:gpr32 = MOVi32imm 10 - ; CHECK-NEXT: STRBBui [[MOVi32imm]], [[COPY]], 0 :: (store (i8) into %ir.dst) ; CHECK-NEXT: [[MOVi32imm1:%[0-9]+]]:gpr32 = MOVi32imm 20 + ; CHECK-NEXT: [[MOVi32imm2:%[0-9]+]]:gpr32 = MOVi32imm 50 + ; CHECK-NEXT: STRBBui [[MOVi32imm]], [[COPY]], 0 :: (store (i8) into %ir.dst) ; CHECK-NEXT: STRBBui [[MOVi32imm1]], [[COPY]], 1 :: (store (i8) into %ir.dst + 1) ; CHECK-NEXT: STRBBui [[MOVi32imm]], [[COPY]], 2 :: (store (i8) into %ir.dst + 2) ; CHECK-NEXT: STRBBui [[MOVi32imm1]], [[COPY]], 3 :: (store (i8) into %ir.dst + 3) - ; CHECK-NEXT: [[MOVi32imm2:%[0-9]+]]:gpr32 = MOVi32imm 50 ; CHECK-NEXT: STRBBui [[MOVi32imm2]], [[COPY]], 4 :: (store (i8) into %ir.dst + 4) ; CHECK-NEXT: STRBBui [[MOVi32imm]], [[COPY]], 5 :: (store (i8) into %ir.dst + 5) ; CHECK-NEXT: STRBBui [[MOVi32imm1]], [[COPY]], 6 :: (store (i8) into %ir.dst + 6) diff --git a/llvm/test/CodeGen/AArch64/O0-pipeline.ll b/llvm/test/CodeGen/AArch64/O0-pipeline.ll index fc068cd2e6cbd..109b0339b3de8 100644 --- a/llvm/test/CodeGen/AArch64/O0-pipeline.ll +++ b/llvm/test/CodeGen/AArch64/O0-pipeline.ll @@ -35,7 +35,6 @@ ; CHECK-NEXT: Analysis containing CSE Info ; CHECK-NEXT: IRTranslator ; CHECK-NEXT: AArch64O0PreLegalizerCombiner -; CHECK-NEXT: Localizer ; CHECK-NEXT: Analysis containing CSE Info ; CHECK-NEXT: Analysis for ComputingKnownBits ; CHECK-NEXT: Legalizer diff --git a/llvm/test/CodeGen/AArch64/O3-pipeline.ll b/llvm/test/CodeGen/AArch64/O3-pipeline.ll index 9a2f7b18e23f1..8d58fe7f3fe0e 100644 --- a/llvm/test/CodeGen/AArch64/O3-pipeline.ll +++ b/llvm/test/CodeGen/AArch64/O3-pipeline.ll @@ -118,7 +118,6 @@ ; CHECK-NEXT: Function Alias Analysis Results ; CHECK-NEXT: IRTranslator ; CHECK-NEXT: AArch64O0PreLegalizerCombiner -; CHECK-NEXT: Localizer ; CHECK-NEXT: Analysis containing CSE Info ; CHECK-NEXT: Analysis for ComputingKnownBits ; CHECK-NEXT: Legalizer diff --git a/llvm/test/CodeGen/AArch64/aarch64-fastcc-stackup.ll b/llvm/test/CodeGen/AArch64/aarch64-fastcc-stackup.ll index 3cfe10ed8a83d..f279485c2a4a6 100644 --- a/llvm/test/CodeGen/AArch64/aarch64-fastcc-stackup.ll +++ b/llvm/test/CodeGen/AArch64/aarch64-fastcc-stackup.ll @@ -24,8 +24,8 @@ define fastcc i64 @baz() { ; CHECK-NEXT: str x30, [sp, #16] // 8-byte Spill ; CHECK-NEXT: .cfi_def_cfa_offset 32 ; CHECK-NEXT: .cfi_offset w30, -16 -; CHECK-NEXT: mov x8, sp ; CHECK-NEXT: mov x7, xzr +; CHECK-NEXT: mov x8, sp ; CHECK-NEXT: str xzr, [x8] ; CHECK-NEXT: mov x0, x7 ; CHECK-NEXT: mov x1, x7 diff --git a/llvm/test/CodeGen/AArch64/aarch64-tbz.ll b/llvm/test/CodeGen/AArch64/aarch64-tbz.ll index f895dabba5714..a6603130899a5 100644 --- a/llvm/test/CodeGen/AArch64/aarch64-tbz.ll +++ b/llvm/test/CodeGen/AArch64/aarch64-tbz.ll @@ -223,17 +223,25 @@ define i32 @tbzfromextract(<8 x i8> %b) { ; ; CHECK-GIO0-LABEL: tbzfromextract: ; CHECK-GIO0: // %bb.0: +; CHECK-GIO0-NEXT: sub sp, sp, #16 +; CHECK-GIO0-NEXT: .cfi_def_cfa_offset 16 ; CHECK-GIO0-NEXT: fmov d1, d0 +; CHECK-GIO0-NEXT: mov w8, wzr +; CHECK-GIO0-NEXT: str w8, [sp, #8] // 4-byte Spill +; CHECK-GIO0-NEXT: mov w8, #1 // =0x1 +; CHECK-GIO0-NEXT: str w8, [sp, #12] // 4-byte Spill ; CHECK-GIO0-NEXT: // implicit-def: $q0 ; CHECK-GIO0-NEXT: fmov d0, d1 ; CHECK-GIO0-NEXT: umov w8, v0.b[0] ; CHECK-GIO0-NEXT: tbz w8, #31, .LBB4_1 ; CHECK-GIO0-NEXT: b .LBB4_2 ; CHECK-GIO0-NEXT: .LBB4_1: // %land.rhs -; CHECK-GIO0-NEXT: mov w0, #1 // =0x1 +; CHECK-GIO0-NEXT: ldr w0, [sp, #12] // 4-byte Reload +; CHECK-GIO0-NEXT: add sp, sp, #16 ; CHECK-GIO0-NEXT: ret ; CHECK-GIO0-NEXT: .LBB4_2: // %land.end -; CHECK-GIO0-NEXT: mov w0, wzr +; CHECK-GIO0-NEXT: ldr w0, [sp, #8] // 4-byte Reload +; CHECK-GIO0-NEXT: add sp, sp, #16 ; CHECK-GIO0-NEXT: ret %e = extractelement <8 x i8> %b, i32 0 %z = zext i8 %e to i32 diff --git a/llvm/test/CodeGen/AArch64/cfguard-checks.ll b/llvm/test/CodeGen/AArch64/cfguard-checks.ll index 0b05b7422fb06..effe4e8d5797f 100644 --- a/llvm/test/CodeGen/AArch64/cfguard-checks.ll +++ b/llvm/test/CodeGen/AArch64/cfguard-checks.ll @@ -40,10 +40,10 @@ entry: ; CHECK-LABEL: func_optnone_cf ; CHECK: adrp x8, target_func ; CHECK-NEXT: add x8, x8, :lo12:target_func + ; CHECK-NEXT: adrp x9, __guard_check_icall_fptr ; CHECK-NEXT: str x8, [sp, #8] ; CHECK-NEXT: ldr x15, [sp, #8] - ; CHECK-NEXT: adrp x8, __guard_check_icall_fptr - ; CHECK-NEXT: ldr x8, [x8, :lo12:__guard_check_icall_fptr] + ; CHECK-NEXT: ldr x8, [x9, :lo12:__guard_check_icall_fptr] ; CHECK-NEXT: blr x8 ; CHECK-NEXT: blr x15 } diff --git a/llvm/test/CodeGen/AArch64/cpa-globalisel.ll b/llvm/test/CodeGen/AArch64/cpa-globalisel.ll index 092c1d189b14a..f5d16dea085bb 100644 --- a/llvm/test/CodeGen/AArch64/cpa-globalisel.ll +++ b/llvm/test/CodeGen/AArch64/cpa-globalisel.ll @@ -51,16 +51,16 @@ entry: define void @maddpt1(i32 %pos, ptr %val) { ; CHECK-CPA-O0-LABEL: maddpt1: ; CHECK-CPA-O0: // %bb.0: // %entry +; CHECK-CPA-O0-NEXT: adrp x10, array2 +; CHECK-CPA-O0-NEXT: add x10, x10, :lo12:array2 +; CHECK-CPA-O0-NEXT: mov w8, #48 // =0x30 +; CHECK-CPA-O0-NEXT: mov w2, w8 ; CHECK-CPA-O0-NEXT: // implicit-def: $x8 ; CHECK-CPA-O0-NEXT: mov w8, w0 ; CHECK-CPA-O0-NEXT: sxtw x8, w8 ; CHECK-CPA-O0-NEXT: mov w9, #48 // =0x30 ; CHECK-CPA-O0-NEXT: // kill: def $x9 killed $w9 -; CHECK-CPA-O0-NEXT: adrp x10, array2 -; CHECK-CPA-O0-NEXT: add x10, x10, :lo12:array2 ; CHECK-CPA-O0-NEXT: maddpt x0, x8, x9, x10 -; CHECK-CPA-O0-NEXT: mov w8, #48 // =0x30 -; CHECK-CPA-O0-NEXT: mov w2, w8 ; CHECK-CPA-O0-NEXT: b memcpy ; ; CHECK-CPA-O3-LABEL: maddpt1: @@ -84,9 +84,9 @@ define void @maddpt1(i32 %pos, ptr %val) { ; CHECK-NOCPA-O0-NEXT: adrp x9, array2 ; CHECK-NOCPA-O0-NEXT: add x9, x9, :lo12:array2 ; CHECK-NOCPA-O0-NEXT: mov w8, #48 // =0x30 -; CHECK-NOCPA-O0-NEXT: smaddl x0, w0, w8, x9 -; CHECK-NOCPA-O0-NEXT: mov w8, #48 // =0x30 ; CHECK-NOCPA-O0-NEXT: mov w2, w8 +; CHECK-NOCPA-O0-NEXT: mov w8, #48 // =0x30 +; CHECK-NOCPA-O0-NEXT: smaddl x0, w0, w8, x9 ; CHECK-NOCPA-O0-NEXT: b memcpy ; ; CHECK-NOCPA-O3-LABEL: maddpt1: @@ -112,22 +112,22 @@ entry: define void @msubpt1(i32 %index, i32 %elem) { ; CHECK-CPA-O0-LABEL: msubpt1: ; CHECK-CPA-O0: // %bb.0: // %entry -; CHECK-CPA-O0-NEXT: // implicit-def: $x8 -; CHECK-CPA-O0-NEXT: mov w8, w0 -; CHECK-CPA-O0-NEXT: sxtw x10, w8 -; CHECK-CPA-O0-NEXT: mov w8, #48 // =0x30 -; CHECK-CPA-O0-NEXT: mov w9, w8 -; CHECK-CPA-O0-NEXT: mov w8, #288 // =0x120 -; CHECK-CPA-O0-NEXT: mov w11, w8 ; CHECK-CPA-O0-NEXT: adrp x8, array2 ; CHECK-CPA-O0-NEXT: add x8, x8, :lo12:array2 -; CHECK-CPA-O0-NEXT: addpt x11, x8, x11 -; CHECK-CPA-O0-NEXT: msubpt x0, x9, x10, x11 -; CHECK-CPA-O0-NEXT: mov w9, #48 // =0x30 -; CHECK-CPA-O0-NEXT: mov w2, w9 +; CHECK-CPA-O0-NEXT: mov w9, #288 // =0x120 +; CHECK-CPA-O0-NEXT: // kill: def $x9 killed $w9 +; CHECK-CPA-O0-NEXT: addpt x10, x8, x9 ; CHECK-CPA-O0-NEXT: mov w9, #96 // =0x60 ; CHECK-CPA-O0-NEXT: // kill: def $x9 killed $w9 ; CHECK-CPA-O0-NEXT: addpt x1, x8, x9 +; CHECK-CPA-O0-NEXT: mov w8, #48 // =0x30 +; CHECK-CPA-O0-NEXT: mov w2, w8 +; CHECK-CPA-O0-NEXT: // implicit-def: $x8 +; CHECK-CPA-O0-NEXT: mov w8, w0 +; CHECK-CPA-O0-NEXT: sxtw x9, w8 +; CHECK-CPA-O0-NEXT: mov w8, #48 // =0x30 +; CHECK-CPA-O0-NEXT: // kill: def $x8 killed $w8 +; CHECK-CPA-O0-NEXT: msubpt x0, x8, x9, x10 ; CHECK-CPA-O0-NEXT: b memcpy ; ; CHECK-CPA-O3-LABEL: msubpt1: @@ -148,19 +148,19 @@ define void @msubpt1(i32 %index, i32 %elem) { ; ; CHECK-NOCPA-O0-LABEL: msubpt1: ; CHECK-NOCPA-O0: // %bb.0: // %entry -; CHECK-NOCPA-O0-NEXT: // implicit-def: $x8 -; CHECK-NOCPA-O0-NEXT: mov w8, w0 -; CHECK-NOCPA-O0-NEXT: sxtw x8, w8 -; CHECK-NOCPA-O0-NEXT: mov w9, #48 // =0x30 -; CHECK-NOCPA-O0-NEXT: // kill: def $x9 killed $w9 -; CHECK-NOCPA-O0-NEXT: mneg x10, x8, x9 -; CHECK-NOCPA-O0-NEXT: adrp x8, array2 -; CHECK-NOCPA-O0-NEXT: add x8, x8, :lo12:array2 -; CHECK-NOCPA-O0-NEXT: add x9, x8, #288 -; CHECK-NOCPA-O0-NEXT: add x0, x9, x10 +; CHECK-NOCPA-O0-NEXT: adrp x9, array2 +; CHECK-NOCPA-O0-NEXT: add x9, x9, :lo12:array2 +; CHECK-NOCPA-O0-NEXT: add x8, x9, #288 +; CHECK-NOCPA-O0-NEXT: add x1, x9, #96 ; CHECK-NOCPA-O0-NEXT: mov w9, #48 // =0x30 ; CHECK-NOCPA-O0-NEXT: mov w2, w9 -; CHECK-NOCPA-O0-NEXT: add x1, x8, #96 +; CHECK-NOCPA-O0-NEXT: // implicit-def: $x9 +; CHECK-NOCPA-O0-NEXT: mov w9, w0 +; CHECK-NOCPA-O0-NEXT: sxtw x9, w9 +; CHECK-NOCPA-O0-NEXT: mov w10, #48 // =0x30 +; CHECK-NOCPA-O0-NEXT: // kill: def $x10 killed $w10 +; CHECK-NOCPA-O0-NEXT: mneg x9, x9, x10 +; CHECK-NOCPA-O0-NEXT: add x0, x8, x9 ; CHECK-NOCPA-O0-NEXT: b memcpy ; ; CHECK-NOCPA-O3-LABEL: msubpt1: @@ -190,14 +190,14 @@ entry: define void @subpt1(i32 %index, i32 %elem) { ; CHECK-CPA-O0-LABEL: subpt1: ; CHECK-CPA-O0: // %bb.0: // %entry -; CHECK-CPA-O0-NEXT: mov w8, #-16 // =0xfffffff0 -; CHECK-CPA-O0-NEXT: smull x9, w0, w8 -; CHECK-CPA-O0-NEXT: adrp x8, array -; CHECK-CPA-O0-NEXT: add x8, x8, :lo12:array -; CHECK-CPA-O0-NEXT: ldr q0, [x8, #32] -; CHECK-CPA-O0-NEXT: mov w10, #96 // =0x60 -; CHECK-CPA-O0-NEXT: // kill: def $x10 killed $w10 -; CHECK-CPA-O0-NEXT: addpt x8, x8, x10 +; CHECK-CPA-O0-NEXT: adrp x10, array +; CHECK-CPA-O0-NEXT: add x10, x10, :lo12:array +; CHECK-CPA-O0-NEXT: mov w8, #96 // =0x60 +; CHECK-CPA-O0-NEXT: // kill: def $x8 killed $w8 +; CHECK-CPA-O0-NEXT: addpt x8, x10, x8 +; CHECK-CPA-O0-NEXT: mov w9, #-16 // =0xfffffff0 +; CHECK-CPA-O0-NEXT: smull x9, w0, w9 +; CHECK-CPA-O0-NEXT: ldr q0, [x10, #32] ; CHECK-CPA-O0-NEXT: str q0, [x8, x9, lsl #4] ; CHECK-CPA-O0-NEXT: ret ; @@ -214,12 +214,12 @@ define void @subpt1(i32 %index, i32 %elem) { ; ; CHECK-NOCPA-O0-LABEL: subpt1: ; CHECK-NOCPA-O0: // %bb.0: // %entry -; CHECK-NOCPA-O0-NEXT: mov w8, #-16 // =0xfffffff0 -; CHECK-NOCPA-O0-NEXT: smull x9, w0, w8 -; CHECK-NOCPA-O0-NEXT: adrp x8, array -; CHECK-NOCPA-O0-NEXT: add x8, x8, :lo12:array -; CHECK-NOCPA-O0-NEXT: ldr q0, [x8, #32] -; CHECK-NOCPA-O0-NEXT: add x8, x8, #96 +; CHECK-NOCPA-O0-NEXT: adrp x10, array +; CHECK-NOCPA-O0-NEXT: add x10, x10, :lo12:array +; CHECK-NOCPA-O0-NEXT: add x8, x10, #96 +; CHECK-NOCPA-O0-NEXT: mov w9, #-16 // =0xfffffff0 +; CHECK-NOCPA-O0-NEXT: smull x9, w0, w9 +; CHECK-NOCPA-O0-NEXT: ldr q0, [x10, #32] ; CHECK-NOCPA-O0-NEXT: str q0, [x8, x9, lsl #4] ; CHECK-NOCPA-O0-NEXT: ret ; @@ -244,14 +244,14 @@ entry: define void @subpt2(i32 %index, i32 %elem) { ; CHECK-CPA-O0-LABEL: subpt2: ; CHECK-CPA-O0: // %bb.0: // %entry -; CHECK-CPA-O0-NEXT: mov x8, xzr -; CHECK-CPA-O0-NEXT: subs x9, x8, w0, sxtw -; CHECK-CPA-O0-NEXT: adrp x8, array -; CHECK-CPA-O0-NEXT: add x8, x8, :lo12:array -; CHECK-CPA-O0-NEXT: ldr q0, [x8, #32] -; CHECK-CPA-O0-NEXT: mov w10, #96 // =0x60 -; CHECK-CPA-O0-NEXT: // kill: def $x10 killed $w10 -; CHECK-CPA-O0-NEXT: addpt x8, x8, x10 +; CHECK-CPA-O0-NEXT: mov x9, xzr +; CHECK-CPA-O0-NEXT: adrp x10, array +; CHECK-CPA-O0-NEXT: add x10, x10, :lo12:array +; CHECK-CPA-O0-NEXT: mov w8, #96 // =0x60 +; CHECK-CPA-O0-NEXT: // kill: def $x8 killed $w8 +; CHECK-CPA-O0-NEXT: addpt x8, x10, x8 +; CHECK-CPA-O0-NEXT: subs x9, x9, w0, sxtw +; CHECK-CPA-O0-NEXT: ldr q0, [x10, #32] ; CHECK-CPA-O0-NEXT: str q0, [x8, x9, lsl #4] ; CHECK-CPA-O0-NEXT: ret ; @@ -268,12 +268,12 @@ define void @subpt2(i32 %index, i32 %elem) { ; ; CHECK-NOCPA-O0-LABEL: subpt2: ; CHECK-NOCPA-O0: // %bb.0: // %entry -; CHECK-NOCPA-O0-NEXT: mov x8, xzr -; CHECK-NOCPA-O0-NEXT: subs x9, x8, w0, sxtw -; CHECK-NOCPA-O0-NEXT: adrp x8, array -; CHECK-NOCPA-O0-NEXT: add x8, x8, :lo12:array -; CHECK-NOCPA-O0-NEXT: ldr q0, [x8, #32] -; CHECK-NOCPA-O0-NEXT: add x8, x8, #96 +; CHECK-NOCPA-O0-NEXT: mov x9, xzr +; CHECK-NOCPA-O0-NEXT: adrp x10, array +; CHECK-NOCPA-O0-NEXT: add x10, x10, :lo12:array +; CHECK-NOCPA-O0-NEXT: add x8, x10, #96 +; CHECK-NOCPA-O0-NEXT: subs x9, x9, w0, sxtw +; CHECK-NOCPA-O0-NEXT: ldr q0, [x10, #32] ; CHECK-NOCPA-O0-NEXT: str q0, [x8, x9, lsl #4] ; CHECK-NOCPA-O0-NEXT: ret ; @@ -645,31 +645,35 @@ entry: define hidden void @multidim() { ; CHECK-CPA-O0-LABEL: multidim: ; CHECK-CPA-O0: // %bb.0: // %entry -; CHECK-CPA-O0-NEXT: str x30, [sp, #-16]! // 8-byte Folded Spill -; CHECK-CPA-O0-NEXT: .cfi_def_cfa_offset 16 +; CHECK-CPA-O0-NEXT: sub sp, sp, #32 +; CHECK-CPA-O0-NEXT: str x30, [sp, #16] // 8-byte Spill +; CHECK-CPA-O0-NEXT: .cfi_def_cfa_offset 32 ; CHECK-CPA-O0-NEXT: .cfi_offset w30, -16 -; CHECK-CPA-O0-NEXT: adrp x8, b -; CHECK-CPA-O0-NEXT: ldrh w9, [x8, :lo12:b] -; CHECK-CPA-O0-NEXT: mov w10, w9 -; CHECK-CPA-O0-NEXT: ldrh w8, [x8, :lo12:b] -; CHECK-CPA-O0-NEXT: add w9, w8, #1 -; CHECK-CPA-O0-NEXT: mov w8, #2 // =0x2 -; CHECK-CPA-O0-NEXT: mov w11, w8 +; CHECK-CPA-O0-NEXT: adrp x9, b ; CHECK-CPA-O0-NEXT: adrp x8, a ; CHECK-CPA-O0-NEXT: add x8, x8, :lo12:a -; CHECK-CPA-O0-NEXT: addpt x8, x8, x11 +; CHECK-CPA-O0-NEXT: mov w10, #2 // =0x2 +; CHECK-CPA-O0-NEXT: // kill: def $x10 killed $w10 +; CHECK-CPA-O0-NEXT: addpt x8, x8, x10 +; CHECK-CPA-O0-NEXT: adrp x10, .L.str +; CHECK-CPA-O0-NEXT: add x10, x10, :lo12:.L.str +; CHECK-CPA-O0-NEXT: str x10, [sp, #8] // 8-byte Spill +; CHECK-CPA-O0-NEXT: ldrh w10, [x9, :lo12:b] +; CHECK-CPA-O0-NEXT: // kill: def $x10 killed $w10 ; CHECK-CPA-O0-NEXT: addpt x8, x8, x10, lsl #1 +; CHECK-CPA-O0-NEXT: ldrh w9, [x9, :lo12:b] +; CHECK-CPA-O0-NEXT: add w9, w9, #1 ; CHECK-CPA-O0-NEXT: ldrb w8, [x8, w9, sxtw] ; CHECK-CPA-O0-NEXT: uxtb w8, w8 ; CHECK-CPA-O0-NEXT: cbnz w8, .LBB14_1 ; CHECK-CPA-O0-NEXT: b .LBB14_2 ; CHECK-CPA-O0-NEXT: .LBB14_1: -; CHECK-CPA-O0-NEXT: adrp x0, .L.str -; CHECK-CPA-O0-NEXT: add x0, x0, :lo12:.L.str +; CHECK-CPA-O0-NEXT: ldr x0, [sp, #8] // 8-byte Reload ; CHECK-CPA-O0-NEXT: bl printf ; CHECK-CPA-O0-NEXT: b .LBB14_2 ; CHECK-CPA-O0-NEXT: .LBB14_2: -; CHECK-CPA-O0-NEXT: ldr x30, [sp], #16 // 8-byte Folded Reload +; CHECK-CPA-O0-NEXT: ldr x30, [sp, #16] // 8-byte Reload +; CHECK-CPA-O0-NEXT: add sp, sp, #32 ; CHECK-CPA-O0-NEXT: ret ; ; CHECK-CPA-O3-LABEL: multidim: @@ -696,29 +700,33 @@ define hidden void @multidim() { ; ; CHECK-NOCPA-O0-LABEL: multidim: ; CHECK-NOCPA-O0: // %bb.0: // %entry -; CHECK-NOCPA-O0-NEXT: str x30, [sp, #-16]! // 8-byte Folded Spill -; CHECK-NOCPA-O0-NEXT: .cfi_def_cfa_offset 16 +; CHECK-NOCPA-O0-NEXT: sub sp, sp, #32 +; CHECK-NOCPA-O0-NEXT: str x30, [sp, #16] // 8-byte Spill +; CHECK-NOCPA-O0-NEXT: .cfi_def_cfa_offset 32 ; CHECK-NOCPA-O0-NEXT: .cfi_offset w30, -16 -; CHECK-NOCPA-O0-NEXT: adrp x8, b -; CHECK-NOCPA-O0-NEXT: ldrh w9, [x8, :lo12:b] -; CHECK-NOCPA-O0-NEXT: mov w10, w9 -; CHECK-NOCPA-O0-NEXT: ldrh w8, [x8, :lo12:b] -; CHECK-NOCPA-O0-NEXT: add w9, w8, #1 +; CHECK-NOCPA-O0-NEXT: adrp x9, b ; CHECK-NOCPA-O0-NEXT: adrp x8, a ; CHECK-NOCPA-O0-NEXT: add x8, x8, :lo12:a ; CHECK-NOCPA-O0-NEXT: add x8, x8, #2 +; CHECK-NOCPA-O0-NEXT: adrp x10, .L.str +; CHECK-NOCPA-O0-NEXT: add x10, x10, :lo12:.L.str +; CHECK-NOCPA-O0-NEXT: str x10, [sp, #8] // 8-byte Spill +; CHECK-NOCPA-O0-NEXT: ldrh w10, [x9, :lo12:b] +; CHECK-NOCPA-O0-NEXT: // kill: def $x10 killed $w10 ; CHECK-NOCPA-O0-NEXT: add x8, x8, x10, lsl #1 +; CHECK-NOCPA-O0-NEXT: ldrh w9, [x9, :lo12:b] +; CHECK-NOCPA-O0-NEXT: add w9, w9, #1 ; CHECK-NOCPA-O0-NEXT: ldrb w8, [x8, w9, sxtw] ; CHECK-NOCPA-O0-NEXT: uxtb w8, w8 ; CHECK-NOCPA-O0-NEXT: cbnz w8, .LBB14_1 ; CHECK-NOCPA-O0-NEXT: b .LBB14_2 ; CHECK-NOCPA-O0-NEXT: .LBB14_1: -; CHECK-NOCPA-O0-NEXT: adrp x0, .L.str -; CHECK-NOCPA-O0-NEXT: add x0, x0, :lo12:.L.str +; CHECK-NOCPA-O0-NEXT: ldr x0, [sp, #8] // 8-byte Reload ; CHECK-NOCPA-O0-NEXT: bl printf ; CHECK-NOCPA-O0-NEXT: b .LBB14_2 ; CHECK-NOCPA-O0-NEXT: .LBB14_2: -; CHECK-NOCPA-O0-NEXT: ldr x30, [sp], #16 // 8-byte Folded Reload +; CHECK-NOCPA-O0-NEXT: ldr x30, [sp, #16] // 8-byte Reload +; CHECK-NOCPA-O0-NEXT: add sp, sp, #32 ; CHECK-NOCPA-O0-NEXT: ret ; ; CHECK-NOCPA-O3-LABEL: multidim: diff --git a/llvm/test/CodeGen/AArch64/large-stack.ll b/llvm/test/CodeGen/AArch64/large-stack.ll index a177398e66e50..2f9c6384e7d1b 100644 --- a/llvm/test/CodeGen/AArch64/large-stack.ll +++ b/llvm/test/CodeGen/AArch64/large-stack.ll @@ -32,15 +32,16 @@ attributes #0 = { noinline optnone "frame-pointer"="all" uwtable } ; CHECK-NEXT: sub sp, sp, #[[STACK2:[0-9]+]], lsl #12 ; CHECK-NEXT: sub sp, sp, #[[STACK3:[0-9]+]] ; CHECK: sub x[[INDEX:[0-9]+]], x[[FRAME]], #8 +; CHECK-NEXT: mov x[[ONE:[0-9]+]], #1 +; CHECK-NEXT: adrp x[[STR:[0-9]+]], .L.str +; CHECK-NEXT: add x[[STR]], x[[STR]], :lo12:.L.str +; CHECK-NEXT: add x[[ARRAY:[0-9]+]], sp, #8 ; CHECK-NEXT: str x0, [x[[INDEX]]] ; CHECK-NEXT: ldr x[[VAL1:[0-9]+]], [x[[INDEX]]] -; CHECK-NEXT: add x10, sp, #8 -; CHECK-NEXT: mov x11, #1 // =0x1 -; CHECK-NEXT: str x11, [x10, x9, lsl #3] -; CHECK-NEXT: ldr x8, [x8] -; CHECK-NEXT: ldr x1, [x10, x8, lsl #3] -; CHECK-NEXT: adrp x0, .L.str -; CHECK-NEXT: add x0, x0, :lo12:.L.str +; CHECK-NEXT: str x[[ONE]], [x[[ARRAY]], x[[VAL1]], lsl #3] +; CHECK-NEXT: ldr x[[VAL2:[0-9]+]], [x[[INDEX]]] +; CHECK-NEXT: ldr x1, [x[[ARRAY]], x[[VAL2]], lsl #3] +; CHECK-NEXT: mov x0, x[[STR]] ; CHECK: bl printf ; CHECK-COUNT-128: add sp, sp, #[[STACK1]], lsl #12 ; CHECK-NEXT: add sp, sp, #[[STACK2]], lsl #12 diff --git a/llvm/test/CodeGen/AArch64/phi-dbg.ll b/llvm/test/CodeGen/AArch64/phi-dbg.ll index 4f7c005f8026f..fba69ad7aaf68 100644 --- a/llvm/test/CodeGen/AArch64/phi-dbg.ll +++ b/llvm/test/CodeGen/AArch64/phi-dbg.ll @@ -28,8 +28,9 @@ bb2: call void @llvm.dbg.value(metadata i32 12, i64 0, metadata !15, metadata !13), !dbg !16 br label %bb3, !dbg !20 -; CHECK: bb.2.bb2: +; CHECK: bb.1.entry: ; CHECK: [[REG0:%[0-9]+]]:gpr32 = MOVi32imm 12 +; CHECK: bb.2.bb2: ; CHECK: [[PHIREG:%[0-9]+]]:gpr32 = COPY [[REG0]] bb3: diff --git a/llvm/test/CodeGen/AArch64/popcount.ll b/llvm/test/CodeGen/AArch64/popcount.ll index bb81776a7f6d6..adbb0dfeb475d 100644 --- a/llvm/test/CodeGen/AArch64/popcount.ll +++ b/llvm/test/CodeGen/AArch64/popcount.ll @@ -768,27 +768,28 @@ Entry: define i32 @ctpop_into_extract(ptr %p) { ; CHECKO0-LABEL: ctpop_into_extract: ; CHECKO0: // %bb.0: -; CHECKO0-NEXT: mov w8, #-1 // =0xffffffff +; CHECKO0-NEXT: mov x8, x0 +; CHECKO0-NEXT: mov w9, #-1 // =0xffffffff ; CHECKO0-NEXT: // implicit-def: $d2 -; CHECKO0-NEXT: fmov s2, w8 -; CHECKO0-NEXT: ldr d0, [x0] +; CHECKO0-NEXT: fmov s2, w9 +; CHECKO0-NEXT: mov w0, wzr +; CHECKO0-NEXT: ldr d0, [x8] ; CHECKO0-NEXT: // implicit-def: $q1 ; CHECKO0-NEXT: fmov d1, d0 -; CHECKO0-NEXT: mov w8, v1.s[0] -; CHECKO0-NEXT: mov w1, w8 +; CHECKO0-NEXT: mov w9, v1.s[0] +; CHECKO0-NEXT: mov w1, w9 ; CHECKO0-NEXT: fmov d1, x1 ; CHECKO0-NEXT: cnt v1.8b, v1.8b ; CHECKO0-NEXT: uaddlv h1, v1.8b ; CHECKO0-NEXT: // kill: def $q1 killed $h1 ; CHECKO0-NEXT: // kill: def $s1 killed $s1 killed $q1 -; CHECKO0-NEXT: fmov w8, s1 +; CHECKO0-NEXT: fmov w9, s1 ; CHECKO0-NEXT: // implicit-def: $q1 ; CHECKO0-NEXT: fmov d1, d2 -; CHECKO0-NEXT: mov v1.s[1], w8 +; CHECKO0-NEXT: mov v1.s[1], w9 ; CHECKO0-NEXT: // kill: def $d1 killed $d1 killed $q1 ; CHECKO0-NEXT: sub v0.2s, v0.2s, v1.2s -; CHECKO0-NEXT: str d0, [x0] -; CHECKO0-NEXT: mov w0, wzr +; CHECKO0-NEXT: str d0, [x8] ; CHECKO0-NEXT: ret ; ; NEON-LABEL: ctpop_into_extract: @@ -868,27 +869,28 @@ define i32 @ctpop_into_extract(ptr %p) { ; ; GISELO0-LABEL: ctpop_into_extract: ; GISELO0: // %bb.0: -; GISELO0-NEXT: mov w8, #-1 // =0xffffffff +; GISELO0-NEXT: mov x8, x0 +; GISELO0-NEXT: mov w9, #-1 // =0xffffffff ; GISELO0-NEXT: // implicit-def: $d2 -; GISELO0-NEXT: fmov s2, w8 -; GISELO0-NEXT: ldr d0, [x0] +; GISELO0-NEXT: fmov s2, w9 +; GISELO0-NEXT: mov w0, wzr +; GISELO0-NEXT: ldr d0, [x8] ; GISELO0-NEXT: // implicit-def: $q1 ; GISELO0-NEXT: fmov d1, d0 -; GISELO0-NEXT: mov w8, v1.s[0] -; GISELO0-NEXT: mov w1, w8 +; GISELO0-NEXT: mov w9, v1.s[0] +; GISELO0-NEXT: mov w1, w9 ; GISELO0-NEXT: fmov d1, x1 ; GISELO0-NEXT: cnt v1.8b, v1.8b ; GISELO0-NEXT: uaddlv h1, v1.8b ; GISELO0-NEXT: // kill: def $q1 killed $h1 ; GISELO0-NEXT: // kill: def $s1 killed $s1 killed $q1 -; GISELO0-NEXT: fmov w8, s1 +; GISELO0-NEXT: fmov w9, s1 ; GISELO0-NEXT: // implicit-def: $q1 ; GISELO0-NEXT: fmov d1, d2 -; GISELO0-NEXT: mov v1.s[1], w8 +; GISELO0-NEXT: mov v1.s[1], w9 ; GISELO0-NEXT: // kill: def $d1 killed $d1 killed $q1 ; GISELO0-NEXT: sub v0.2s, v0.2s, v1.2s -; GISELO0-NEXT: str d0, [x0] -; GISELO0-NEXT: mov w0, wzr +; GISELO0-NEXT: str d0, [x8] ; GISELO0-NEXT: ret %1 = load <2 x i32>, ptr %p, align 4 %2 = extractelement <2 x i32> %1, i64 0 diff --git a/llvm/test/CodeGen/AArch64/pr92062.ll b/llvm/test/CodeGen/AArch64/pr92062.ll index 6111ee0fbe18f..c6f8922cc2a48 100644 --- a/llvm/test/CodeGen/AArch64/pr92062.ll +++ b/llvm/test/CodeGen/AArch64/pr92062.ll @@ -10,8 +10,8 @@ define void @foo() { ; CHECK: // %bb.0: // %bb ; CHECK-NEXT: adrp x8, :got:p ; CHECK-NEXT: ldr x8, [x8, :got_lo12:p] -; CHECK-NEXT: ldr x8, [x8] ; CHECK-NEXT: mov x9, xzr +; CHECK-NEXT: ldr x8, [x8] ; CHECK-NEXT: str x8, [x9] ; CHECK-NEXT: ret bb: diff --git a/llvm/test/CodeGen/AArch64/vararg.ll b/llvm/test/CodeGen/AArch64/vararg.ll index 291eee2ddf706..33ac571b6d4c1 100644 --- a/llvm/test/CodeGen/AArch64/vararg.ll +++ b/llvm/test/CodeGen/AArch64/vararg.ll @@ -86,6 +86,8 @@ define i64 @vararg(...) #0 { ; CHECK-GI-NEXT: str q5, [sp, #96] ; CHECK-GI-NEXT: str q6, [sp, #112] ; CHECK-GI-NEXT: str q7, [sp, #128] +; CHECK-GI-NEXT: mov w8, #1 // =0x1 +; CHECK-GI-NEXT: mov w0, w8 ; CHECK-GI-NEXT: add x9, sp, #8 ; CHECK-GI-NEXT: add x8, x29, #16 ; CHECK-GI-NEXT: str x8, [x9] @@ -97,8 +99,6 @@ define i64 @vararg(...) #0 { ; CHECK-GI-NEXT: str w8, [x9, #24] ; CHECK-GI-NEXT: mov w8, #-128 // =0xffffff80 ; CHECK-GI-NEXT: str w8, [x9, #28] -; CHECK-GI-NEXT: mov w8, #1 // =0x1 -; CHECK-GI-NEXT: mov w0, w8 ; CHECK-GI-NEXT: .cfi_def_cfa wsp, 224 ; CHECK-GI-NEXT: ldp x29, x30, [sp, #208] // 16-byte Folded Reload ; CHECK-GI-NEXT: add sp, sp, #224 @@ -171,6 +171,8 @@ define i64 @vararg_many_gpr(i64 %a1, i64 %a2, i64 %a3, i64 %a4, i64 %a5, i64 %a6 ; CHECK-GI-NEXT: str q5, [sp, #96] ; CHECK-GI-NEXT: str q6, [sp, #112] ; CHECK-GI-NEXT: str q7, [sp, #128] +; CHECK-GI-NEXT: mov w8, #1 // =0x1 +; CHECK-GI-NEXT: mov w0, w8 ; CHECK-GI-NEXT: add x9, sp, #8 ; CHECK-GI-NEXT: add x8, x29, #16 ; CHECK-GI-NEXT: str x8, [x9] @@ -182,8 +184,6 @@ define i64 @vararg_many_gpr(i64 %a1, i64 %a2, i64 %a3, i64 %a4, i64 %a5, i64 %a6 ; CHECK-GI-NEXT: str w8, [x9, #24] ; CHECK-GI-NEXT: mov w8, #-128 // =0xffffff80 ; CHECK-GI-NEXT: str w8, [x9, #28] -; CHECK-GI-NEXT: mov w8, #1 // =0x1 -; CHECK-GI-NEXT: mov w0, w8 ; CHECK-GI-NEXT: .cfi_def_cfa wsp, 176 ; CHECK-GI-NEXT: ldp x29, x30, [sp, #160] // 16-byte Folded Reload ; CHECK-GI-NEXT: add sp, sp, #176 @@ -256,6 +256,8 @@ define i64 @vararg_many_float(float %a1, float %a2, float %a3, float %a4, float ; CHECK-GI-NEXT: str x6, [sp, #80] ; CHECK-GI-NEXT: str x7, [sp, #88] ; CHECK-GI-NEXT: str q7, [sp, #16] +; CHECK-GI-NEXT: mov w8, #1 // =0x1 +; CHECK-GI-NEXT: mov w0, w8 ; CHECK-GI-NEXT: add x9, sp, #8 ; CHECK-GI-NEXT: add x8, x29, #16 ; CHECK-GI-NEXT: str x8, [x9] @@ -267,8 +269,6 @@ define i64 @vararg_many_float(float %a1, float %a2, float %a3, float %a4, float ; CHECK-GI-NEXT: str w8, [x9, #24] ; CHECK-GI-NEXT: mov w8, #-16 // =0xfffffff0 ; CHECK-GI-NEXT: str w8, [x9, #28] -; CHECK-GI-NEXT: mov w8, #1 // =0x1 -; CHECK-GI-NEXT: mov w0, w8 ; CHECK-GI-NEXT: .cfi_def_cfa wsp, 112 ; CHECK-GI-NEXT: ldp x29, x30, [sp, #96] // 16-byte Folded Reload ; CHECK-GI-NEXT: add sp, sp, #112 @@ -351,6 +351,8 @@ define i64 @gpr1_fpr1(i32 %i, float %f, ...) #0 { ; CHECK-GI-NEXT: str q5, [sp, #80] ; CHECK-GI-NEXT: str q6, [sp, #96] ; CHECK-GI-NEXT: str q7, [sp, #112] +; CHECK-GI-NEXT: mov w8, #1 // =0x1 +; CHECK-GI-NEXT: mov w0, w8 ; CHECK-GI-NEXT: add x9, sp, #8 ; CHECK-GI-NEXT: add x8, x29, #16 ; CHECK-GI-NEXT: str x8, [x9] @@ -362,8 +364,6 @@ define i64 @gpr1_fpr1(i32 %i, float %f, ...) #0 { ; CHECK-GI-NEXT: str w8, [x9, #24] ; CHECK-GI-NEXT: mov w8, #-112 // =0xffffff90 ; CHECK-GI-NEXT: str w8, [x9, #28] -; CHECK-GI-NEXT: mov w8, #1 // =0x1 -; CHECK-GI-NEXT: mov w0, w8 ; CHECK-GI-NEXT: .cfi_def_cfa wsp, 208 ; CHECK-GI-NEXT: ldp x29, x30, [sp, #192] // 16-byte Folded Reload ; CHECK-GI-NEXT: add sp, sp, #208 diff --git a/llvm/test/CodeGen/AArch64/win64_vararg_float.ll b/llvm/test/CodeGen/AArch64/win64_vararg_float.ll index a99b29fa3bef9..d3f9b45c6b687 100644 --- a/llvm/test/CodeGen/AArch64/win64_vararg_float.ll +++ b/llvm/test/CodeGen/AArch64/win64_vararg_float.ll @@ -118,12 +118,12 @@ define void @call_f_va() nounwind { ; ; GISEL-LABEL: call_f_va: ; GISEL: // %bb.0: // %entry -; GISEL-NEXT: fmov s0, #1.00000000 -; GISEL-NEXT: fmov w0, s0 +; GISEL-NEXT: fmov s1, #1.00000000 ; GISEL-NEXT: mov w1, #2 // =0x2 ; GISEL-NEXT: fmov d0, #3.00000000 -; GISEL-NEXT: fmov x2, d0 ; GISEL-NEXT: mov w3, #4 // =0x4 +; GISEL-NEXT: fmov w0, s1 +; GISEL-NEXT: fmov x2, d0 ; GISEL-NEXT: b other_f_va_fn entry: tail call void (float, i32, ...) @other_f_va_fn(float 1.000000e+00, i32 2, double 3.000000e+00, i32 4) @@ -151,12 +151,12 @@ define void @call_d_va() nounwind { ; ; GISEL-LABEL: call_d_va: ; GISEL: // %bb.0: // %entry -; GISEL-NEXT: fmov d0, #1.00000000 -; GISEL-NEXT: fmov x0, d0 +; GISEL-NEXT: fmov d1, #1.00000000 ; GISEL-NEXT: mov w1, #2 // =0x2 ; GISEL-NEXT: fmov d0, #3.00000000 -; GISEL-NEXT: fmov x2, d0 ; GISEL-NEXT: mov w3, #4 // =0x4 +; GISEL-NEXT: fmov x0, d1 +; GISEL-NEXT: fmov x2, d0 ; GISEL-NEXT: b other_d_va_fn entry: tail call void (double, i32, ...) @other_d_va_fn(double 1.000000e+00, i32 2, double 3.000000e+00, i32 4) diff --git a/llvm/test/CodeGen/AArch64/win64_vararg_float_cc.ll b/llvm/test/CodeGen/AArch64/win64_vararg_float_cc.ll index 029d2da43b691..049cee9341417 100644 --- a/llvm/test/CodeGen/AArch64/win64_vararg_float_cc.ll +++ b/llvm/test/CodeGen/AArch64/win64_vararg_float_cc.ll @@ -138,12 +138,12 @@ define void @call_f_va() nounwind { ; GISEL-LABEL: call_f_va: ; GISEL: // %bb.0: // %entry ; GISEL-NEXT: str x30, [sp, #-16]! // 8-byte Folded Spill -; GISEL-NEXT: fmov s0, #1.00000000 -; GISEL-NEXT: fmov w0, s0 +; GISEL-NEXT: fmov s1, #1.00000000 ; GISEL-NEXT: mov w1, #2 // =0x2 ; GISEL-NEXT: fmov d0, #3.00000000 -; GISEL-NEXT: fmov x2, d0 ; GISEL-NEXT: mov w3, #4 // =0x4 +; GISEL-NEXT: fmov w0, s1 +; GISEL-NEXT: fmov x2, d0 ; GISEL-NEXT: bl other_f_va_fn ; GISEL-NEXT: ldr x30, [sp], #16 // 8-byte Folded Reload ; GISEL-NEXT: ret @@ -180,12 +180,12 @@ define void @call_d_va() nounwind { ; GISEL-LABEL: call_d_va: ; GISEL: // %bb.0: // %entry ; GISEL-NEXT: str x30, [sp, #-16]! // 8-byte Folded Spill -; GISEL-NEXT: fmov d0, #1.00000000 -; GISEL-NEXT: fmov x0, d0 +; GISEL-NEXT: fmov d1, #1.00000000 ; GISEL-NEXT: mov w1, #2 // =0x2 ; GISEL-NEXT: fmov d0, #3.00000000 -; GISEL-NEXT: fmov x2, d0 ; GISEL-NEXT: mov w3, #4 // =0x4 +; GISEL-NEXT: fmov x0, d1 +; GISEL-NEXT: fmov x2, d0 ; GISEL-NEXT: bl other_d_va_fn ; GISEL-NEXT: ldr x30, [sp], #16 // 8-byte Folded Reload ; GISEL-NEXT: ret diff --git a/llvm/test/Other/print-changed-machine.ll b/llvm/test/Other/print-changed-machine.ll index 7b2b9b3c10d95..e2143d113a7f1 100644 --- a/llvm/test/Other/print-changed-machine.ll +++ b/llvm/test/Other/print-changed-machine.ll @@ -24,7 +24,7 @@ ; QUIET: *** IR Dump After IRTranslator (ir-translator) on foo *** ; QUIET-NOT: *** -; QUIET: *** IR Dump After Localizer (localizer) on foo *** +; QUIET: *** IR Dump After Legalizer (legalizer) on foo *** ; RUN: llc -filetype=null -mtriple=aarch64 -O0 -print-changed -filter-passes=ir-translator,legalizer %s 2>&1 | \ ; RUN: FileCheck %s --check-prefixes=VERBOSE-FILTER _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
