aboutsummaryrefslogtreecommitdiff
path: root/llvm/test/CodeGen/AArch64/sme-disable-gisel-fisel.ll
diff options
context:
space:
mode:
Diffstat (limited to 'llvm/test/CodeGen/AArch64/sme-disable-gisel-fisel.ll')
-rw-r--r--llvm/test/CodeGen/AArch64/sme-disable-gisel-fisel.ll79
1 files changed, 55 insertions, 24 deletions
diff --git a/llvm/test/CodeGen/AArch64/sme-disable-gisel-fisel.ll b/llvm/test/CodeGen/AArch64/sme-disable-gisel-fisel.ll
index 254e37e..d786ffd 100644
--- a/llvm/test/CodeGen/AArch64/sme-disable-gisel-fisel.ll
+++ b/llvm/test/CodeGen/AArch64/sme-disable-gisel-fisel.ll
@@ -1,7 +1,7 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc -fast-isel=true -global-isel=false -fast-isel-abort=0 -mtriple=aarch64-linux-gnu -mattr=+sme2 < %s \
+; RUN: llc -fast-isel=true -global-isel=false -fast-isel-abort=0 -mtriple=aarch64-linux-gnu -mattr=+sve -mattr=+sme2 < %s \
; RUN: | FileCheck %s --check-prefixes=CHECK-COMMON,CHECK-FISEL
-; RUN: llc -fast-isel=false -global-isel=true -global-isel-abort=0 -mtriple=aarch64-linux-gnu -mattr=+sme2 < %s \
+; RUN: llc -fast-isel=false -global-isel=true -global-isel-abort=0 -mtriple=aarch64-linux-gnu -mattr=+sve -mattr=+sme2 < %s \
; RUN: | FileCheck %s --check-prefixes=CHECK-COMMON,CHECK-GISEL
@@ -17,6 +17,8 @@ define double @nonstreaming_caller_streaming_callee(double %x) nounwind noinline
; CHECK-FISEL-NEXT: stp d11, d10, [sp, #48] // 16-byte Folded Spill
; CHECK-FISEL-NEXT: stp d9, d8, [sp, #64] // 16-byte Folded Spill
; CHECK-FISEL-NEXT: str x30, [sp, #80] // 8-byte Folded Spill
+; CHECK-FISEL-NEXT: cntd x9
+; CHECK-FISEL-NEXT: str x9, [sp, #88] // 8-byte Folded Spill
; CHECK-FISEL-NEXT: str d0, [sp] // 8-byte Folded Spill
; CHECK-FISEL-NEXT: smstart sm
; CHECK-FISEL-NEXT: ldr d0, [sp] // 8-byte Folded Reload
@@ -43,6 +45,8 @@ define double @nonstreaming_caller_streaming_callee(double %x) nounwind noinline
; CHECK-GISEL-NEXT: stp d11, d10, [sp, #48] // 16-byte Folded Spill
; CHECK-GISEL-NEXT: stp d9, d8, [sp, #64] // 16-byte Folded Spill
; CHECK-GISEL-NEXT: str x30, [sp, #80] // 8-byte Folded Spill
+; CHECK-GISEL-NEXT: cntd x9
+; CHECK-GISEL-NEXT: str x9, [sp, #88] // 8-byte Folded Spill
; CHECK-GISEL-NEXT: str d0, [sp] // 8-byte Folded Spill
; CHECK-GISEL-NEXT: smstart sm
; CHECK-GISEL-NEXT: ldr d0, [sp] // 8-byte Folded Reload
@@ -76,6 +80,8 @@ define double @streaming_caller_nonstreaming_callee(double %x) nounwind noinline
; CHECK-COMMON-NEXT: stp d11, d10, [sp, #48] // 16-byte Folded Spill
; CHECK-COMMON-NEXT: stp d9, d8, [sp, #64] // 16-byte Folded Spill
; CHECK-COMMON-NEXT: str x30, [sp, #80] // 8-byte Folded Spill
+; CHECK-COMMON-NEXT: cntd x9
+; CHECK-COMMON-NEXT: str x9, [sp, #88] // 8-byte Folded Spill
; CHECK-COMMON-NEXT: str d0, [sp] // 8-byte Folded Spill
; CHECK-COMMON-NEXT: smstop sm
; CHECK-COMMON-NEXT: ldr d0, [sp] // 8-byte Folded Reload
@@ -102,12 +108,17 @@ entry:
define double @locally_streaming_caller_normal_callee(double %x) nounwind noinline optnone "aarch64_pstate_sm_body" {
; CHECK-COMMON-LABEL: locally_streaming_caller_normal_callee:
; CHECK-COMMON: // %bb.0:
-; CHECK-COMMON-NEXT: sub sp, sp, #112
+; CHECK-COMMON-NEXT: sub sp, sp, #128
; CHECK-COMMON-NEXT: stp d15, d14, [sp, #32] // 16-byte Folded Spill
; CHECK-COMMON-NEXT: stp d13, d12, [sp, #48] // 16-byte Folded Spill
; CHECK-COMMON-NEXT: stp d11, d10, [sp, #64] // 16-byte Folded Spill
; CHECK-COMMON-NEXT: stp d9, d8, [sp, #80] // 16-byte Folded Spill
; CHECK-COMMON-NEXT: str x30, [sp, #96] // 8-byte Folded Spill
+; CHECK-COMMON-NEXT: rdsvl x9, #1
+; CHECK-COMMON-NEXT: lsr x9, x9, #3
+; CHECK-COMMON-NEXT: str x9, [sp, #104] // 8-byte Folded Spill
+; CHECK-COMMON-NEXT: cntd x9
+; CHECK-COMMON-NEXT: str x9, [sp, #112] // 8-byte Folded Spill
; CHECK-COMMON-NEXT: str d0, [sp, #24] // 8-byte Folded Spill
; CHECK-COMMON-NEXT: smstart sm
; CHECK-COMMON-NEXT: ldr d0, [sp, #24] // 8-byte Folded Reload
@@ -129,7 +140,7 @@ define double @locally_streaming_caller_normal_callee(double %x) nounwind noinli
; CHECK-COMMON-NEXT: ldp d11, d10, [sp, #64] // 16-byte Folded Reload
; CHECK-COMMON-NEXT: ldp d13, d12, [sp, #48] // 16-byte Folded Reload
; CHECK-COMMON-NEXT: ldp d15, d14, [sp, #32] // 16-byte Folded Reload
-; CHECK-COMMON-NEXT: add sp, sp, #112
+; CHECK-COMMON-NEXT: add sp, sp, #128
; CHECK-COMMON-NEXT: ret
%call = call double @normal_callee(double %x);
%add = fadd double %call, 4.200000e+01
@@ -166,11 +177,16 @@ define double @normal_caller_to_locally_streaming_callee(double %x) nounwind noi
define void @locally_streaming_caller_streaming_callee_ptr(ptr %p) nounwind noinline optnone "aarch64_pstate_sm_body" {
; CHECK-COMMON-LABEL: locally_streaming_caller_streaming_callee_ptr:
; CHECK-COMMON: // %bb.0:
-; CHECK-COMMON-NEXT: stp d15, d14, [sp, #-80]! // 16-byte Folded Spill
+; CHECK-COMMON-NEXT: stp d15, d14, [sp, #-96]! // 16-byte Folded Spill
; CHECK-COMMON-NEXT: stp d13, d12, [sp, #16] // 16-byte Folded Spill
; CHECK-COMMON-NEXT: stp d11, d10, [sp, #32] // 16-byte Folded Spill
; CHECK-COMMON-NEXT: stp d9, d8, [sp, #48] // 16-byte Folded Spill
; CHECK-COMMON-NEXT: str x30, [sp, #64] // 8-byte Folded Spill
+; CHECK-COMMON-NEXT: rdsvl x9, #1
+; CHECK-COMMON-NEXT: lsr x9, x9, #3
+; CHECK-COMMON-NEXT: str x9, [sp, #72] // 8-byte Folded Spill
+; CHECK-COMMON-NEXT: cntd x9
+; CHECK-COMMON-NEXT: str x9, [sp, #80] // 8-byte Folded Spill
; CHECK-COMMON-NEXT: smstart sm
; CHECK-COMMON-NEXT: blr x0
; CHECK-COMMON-NEXT: smstop sm
@@ -178,7 +194,7 @@ define void @locally_streaming_caller_streaming_callee_ptr(ptr %p) nounwind noin
; CHECK-COMMON-NEXT: ldp d9, d8, [sp, #48] // 16-byte Folded Reload
; CHECK-COMMON-NEXT: ldp d11, d10, [sp, #32] // 16-byte Folded Reload
; CHECK-COMMON-NEXT: ldp d13, d12, [sp, #16] // 16-byte Folded Reload
-; CHECK-COMMON-NEXT: ldp d15, d14, [sp], #80 // 16-byte Folded Reload
+; CHECK-COMMON-NEXT: ldp d15, d14, [sp], #96 // 16-byte Folded Reload
; CHECK-COMMON-NEXT: ret
call void %p() "aarch64_pstate_sm_enabled"
ret void
@@ -192,6 +208,8 @@ define void @normal_call_to_streaming_callee_ptr(ptr %p) nounwind noinline optno
; CHECK-COMMON-NEXT: stp d11, d10, [sp, #32] // 16-byte Folded Spill
; CHECK-COMMON-NEXT: stp d9, d8, [sp, #48] // 16-byte Folded Spill
; CHECK-COMMON-NEXT: str x30, [sp, #64] // 8-byte Folded Spill
+; CHECK-COMMON-NEXT: cntd x9
+; CHECK-COMMON-NEXT: str x9, [sp, #72] // 8-byte Folded Spill
; CHECK-COMMON-NEXT: smstart sm
; CHECK-COMMON-NEXT: blr x0
; CHECK-COMMON-NEXT: smstop sm
@@ -214,7 +232,8 @@ declare double @za_shared_callee(double) "aarch64_inout_za"
define double @za_new_caller_to_za_shared_callee(double %x) nounwind noinline optnone "aarch64_new_za"{
; CHECK-COMMON-LABEL: za_new_caller_to_za_shared_callee:
; CHECK-COMMON: // %bb.0: // %prelude
-; CHECK-COMMON-NEXT: stp x29, x30, [sp, #-16]! // 16-byte Folded Spill
+; CHECK-COMMON-NEXT: stp x29, x30, [sp, #-32]! // 16-byte Folded Spill
+; CHECK-COMMON-NEXT: str x19, [sp, #16] // 8-byte Folded Spill
; CHECK-COMMON-NEXT: mov x29, sp
; CHECK-COMMON-NEXT: sub sp, sp, #16
; CHECK-COMMON-NEXT: rdsvl x8, #1
@@ -240,7 +259,8 @@ define double @za_new_caller_to_za_shared_callee(double %x) nounwind noinline o
; CHECK-COMMON-NEXT: fadd d0, d0, d1
; CHECK-COMMON-NEXT: smstop za
; CHECK-COMMON-NEXT: mov sp, x29
-; CHECK-COMMON-NEXT: ldp x29, x30, [sp], #16 // 16-byte Folded Reload
+; CHECK-COMMON-NEXT: ldr x19, [sp, #16] // 8-byte Folded Reload
+; CHECK-COMMON-NEXT: ldp x29, x30, [sp], #32 // 16-byte Folded Reload
; CHECK-COMMON-NEXT: ret
entry:
%call = call double @za_shared_callee(double %x)
@@ -251,7 +271,8 @@ entry:
define double @za_shared_caller_to_za_none_callee(double %x) nounwind noinline optnone "aarch64_inout_za"{
; CHECK-COMMON-LABEL: za_shared_caller_to_za_none_callee:
; CHECK-COMMON: // %bb.0: // %entry
-; CHECK-COMMON-NEXT: stp x29, x30, [sp, #-16]! // 16-byte Folded Spill
+; CHECK-COMMON-NEXT: stp x29, x30, [sp, #-32]! // 16-byte Folded Spill
+; CHECK-COMMON-NEXT: str x19, [sp, #16] // 8-byte Folded Spill
; CHECK-COMMON-NEXT: mov x29, sp
; CHECK-COMMON-NEXT: sub sp, sp, #16
; CHECK-COMMON-NEXT: rdsvl x8, #1
@@ -279,7 +300,8 @@ define double @za_shared_caller_to_za_none_callee(double %x) nounwind noinline
; CHECK-COMMON-NEXT: fmov d1, x8
; CHECK-COMMON-NEXT: fadd d0, d0, d1
; CHECK-COMMON-NEXT: mov sp, x29
-; CHECK-COMMON-NEXT: ldp x29, x30, [sp], #16 // 16-byte Folded Reload
+; CHECK-COMMON-NEXT: ldr x19, [sp, #16] // 8-byte Folded Reload
+; CHECK-COMMON-NEXT: ldp x29, x30, [sp], #32 // 16-byte Folded Reload
; CHECK-COMMON-NEXT: ret
entry:
%call = call double @normal_callee(double %x)
@@ -291,7 +313,8 @@ entry:
define fp128 @f128_call_za(fp128 %a, fp128 %b) "aarch64_inout_za" nounwind {
; CHECK-COMMON-LABEL: f128_call_za:
; CHECK-COMMON: // %bb.0:
-; CHECK-COMMON-NEXT: stp x29, x30, [sp, #-16]! // 16-byte Folded Spill
+; CHECK-COMMON-NEXT: stp x29, x30, [sp, #-32]! // 16-byte Folded Spill
+; CHECK-COMMON-NEXT: str x19, [sp, #16] // 8-byte Folded Spill
; CHECK-COMMON-NEXT: mov x29, sp
; CHECK-COMMON-NEXT: sub sp, sp, #16
; CHECK-COMMON-NEXT: rdsvl x8, #1
@@ -314,7 +337,8 @@ define fp128 @f128_call_za(fp128 %a, fp128 %b) "aarch64_inout_za" nounwind {
; CHECK-COMMON-NEXT: .LBB8_2:
; CHECK-COMMON-NEXT: msr TPIDR2_EL0, xzr
; CHECK-COMMON-NEXT: mov sp, x29
-; CHECK-COMMON-NEXT: ldp x29, x30, [sp], #16 // 16-byte Folded Reload
+; CHECK-COMMON-NEXT: ldr x19, [sp, #16] // 8-byte Folded Reload
+; CHECK-COMMON-NEXT: ldp x29, x30, [sp], #32 // 16-byte Folded Reload
; CHECK-COMMON-NEXT: ret
%res = fadd fp128 %a, %b
ret fp128 %res
@@ -326,21 +350,22 @@ define fp128 @f128_call_sm(fp128 %a, fp128 %b) "aarch64_pstate_sm_enabled" nounw
; CHECK-COMMON-LABEL: f128_call_sm:
; CHECK-COMMON: // %bb.0:
; CHECK-COMMON-NEXT: sub sp, sp, #112
+; CHECK-COMMON-NEXT: cntd x9
; CHECK-COMMON-NEXT: stp d15, d14, [sp, #32] // 16-byte Folded Spill
; CHECK-COMMON-NEXT: stp d13, d12, [sp, #48] // 16-byte Folded Spill
; CHECK-COMMON-NEXT: stp d11, d10, [sp, #64] // 16-byte Folded Spill
; CHECK-COMMON-NEXT: stp d9, d8, [sp, #80] // 16-byte Folded Spill
-; CHECK-COMMON-NEXT: str x30, [sp, #96] // 8-byte Folded Spill
+; CHECK-COMMON-NEXT: stp x30, x9, [sp, #96] // 16-byte Folded Spill
; CHECK-COMMON-NEXT: stp q0, q1, [sp] // 32-byte Folded Spill
; CHECK-COMMON-NEXT: smstop sm
; CHECK-COMMON-NEXT: ldp q0, q1, [sp] // 32-byte Folded Reload
; CHECK-COMMON-NEXT: bl __addtf3
; CHECK-COMMON-NEXT: str q0, [sp, #16] // 16-byte Folded Spill
; CHECK-COMMON-NEXT: smstart sm
-; CHECK-COMMON-NEXT: ldp d9, d8, [sp, #80] // 16-byte Folded Reload
; CHECK-COMMON-NEXT: ldr q0, [sp, #16] // 16-byte Folded Reload
-; CHECK-COMMON-NEXT: ldp d11, d10, [sp, #64] // 16-byte Folded Reload
+; CHECK-COMMON-NEXT: ldp d9, d8, [sp, #80] // 16-byte Folded Reload
; CHECK-COMMON-NEXT: ldr x30, [sp, #96] // 8-byte Folded Reload
+; CHECK-COMMON-NEXT: ldp d11, d10, [sp, #64] // 16-byte Folded Reload
; CHECK-COMMON-NEXT: ldp d13, d12, [sp, #48] // 16-byte Folded Reload
; CHECK-COMMON-NEXT: ldp d15, d14, [sp, #32] // 16-byte Folded Reload
; CHECK-COMMON-NEXT: add sp, sp, #112
@@ -353,7 +378,8 @@ define fp128 @f128_call_sm(fp128 %a, fp128 %b) "aarch64_pstate_sm_enabled" nounw
define double @frem_call_za(double %a, double %b) "aarch64_inout_za" nounwind {
; CHECK-COMMON-LABEL: frem_call_za:
; CHECK-COMMON: // %bb.0:
-; CHECK-COMMON-NEXT: stp x29, x30, [sp, #-16]! // 16-byte Folded Spill
+; CHECK-COMMON-NEXT: stp x29, x30, [sp, #-32]! // 16-byte Folded Spill
+; CHECK-COMMON-NEXT: str x19, [sp, #16] // 8-byte Folded Spill
; CHECK-COMMON-NEXT: mov x29, sp
; CHECK-COMMON-NEXT: sub sp, sp, #16
; CHECK-COMMON-NEXT: rdsvl x8, #1
@@ -376,7 +402,8 @@ define double @frem_call_za(double %a, double %b) "aarch64_inout_za" nounwind {
; CHECK-COMMON-NEXT: .LBB10_2:
; CHECK-COMMON-NEXT: msr TPIDR2_EL0, xzr
; CHECK-COMMON-NEXT: mov sp, x29
-; CHECK-COMMON-NEXT: ldp x29, x30, [sp], #16 // 16-byte Folded Reload
+; CHECK-COMMON-NEXT: ldr x19, [sp, #16] // 8-byte Folded Reload
+; CHECK-COMMON-NEXT: ldp x29, x30, [sp], #32 // 16-byte Folded Reload
; CHECK-COMMON-NEXT: ret
%res = frem double %a, %b
ret double %res
@@ -387,21 +414,22 @@ define float @frem_call_sm(float %a, float %b) "aarch64_pstate_sm_enabled" nounw
; CHECK-COMMON-LABEL: frem_call_sm:
; CHECK-COMMON: // %bb.0:
; CHECK-COMMON-NEXT: sub sp, sp, #96
+; CHECK-COMMON-NEXT: cntd x9
; CHECK-COMMON-NEXT: stp d15, d14, [sp, #16] // 16-byte Folded Spill
; CHECK-COMMON-NEXT: stp d13, d12, [sp, #32] // 16-byte Folded Spill
; CHECK-COMMON-NEXT: stp d11, d10, [sp, #48] // 16-byte Folded Spill
; CHECK-COMMON-NEXT: stp d9, d8, [sp, #64] // 16-byte Folded Spill
-; CHECK-COMMON-NEXT: str x30, [sp, #80] // 8-byte Folded Spill
+; CHECK-COMMON-NEXT: stp x30, x9, [sp, #80] // 16-byte Folded Spill
; CHECK-COMMON-NEXT: stp s0, s1, [sp, #8] // 8-byte Folded Spill
; CHECK-COMMON-NEXT: smstop sm
; CHECK-COMMON-NEXT: ldp s0, s1, [sp, #8] // 8-byte Folded Reload
; CHECK-COMMON-NEXT: bl fmodf
; CHECK-COMMON-NEXT: str s0, [sp, #12] // 4-byte Folded Spill
; CHECK-COMMON-NEXT: smstart sm
-; CHECK-COMMON-NEXT: ldp d9, d8, [sp, #64] // 16-byte Folded Reload
; CHECK-COMMON-NEXT: ldr s0, [sp, #12] // 4-byte Folded Reload
-; CHECK-COMMON-NEXT: ldp d11, d10, [sp, #48] // 16-byte Folded Reload
+; CHECK-COMMON-NEXT: ldp d9, d8, [sp, #64] // 16-byte Folded Reload
; CHECK-COMMON-NEXT: ldr x30, [sp, #80] // 8-byte Folded Reload
+; CHECK-COMMON-NEXT: ldp d11, d10, [sp, #48] // 16-byte Folded Reload
; CHECK-COMMON-NEXT: ldp d13, d12, [sp, #32] // 16-byte Folded Reload
; CHECK-COMMON-NEXT: ldp d15, d14, [sp, #16] // 16-byte Folded Reload
; CHECK-COMMON-NEXT: add sp, sp, #96
@@ -414,12 +442,14 @@ define float @frem_call_sm(float %a, float %b) "aarch64_pstate_sm_enabled" nounw
define float @frem_call_sm_compat(float %a, float %b) "aarch64_pstate_sm_compatible" nounwind {
; CHECK-COMMON-LABEL: frem_call_sm_compat:
; CHECK-COMMON: // %bb.0:
-; CHECK-COMMON-NEXT: sub sp, sp, #96
+; CHECK-COMMON-NEXT: sub sp, sp, #112
+; CHECK-COMMON-NEXT: cntd x9
; CHECK-COMMON-NEXT: stp d15, d14, [sp, #16] // 16-byte Folded Spill
; CHECK-COMMON-NEXT: stp d13, d12, [sp, #32] // 16-byte Folded Spill
; CHECK-COMMON-NEXT: stp d11, d10, [sp, #48] // 16-byte Folded Spill
; CHECK-COMMON-NEXT: stp d9, d8, [sp, #64] // 16-byte Folded Spill
-; CHECK-COMMON-NEXT: stp x30, x19, [sp, #80] // 16-byte Folded Spill
+; CHECK-COMMON-NEXT: stp x30, x9, [sp, #80] // 16-byte Folded Spill
+; CHECK-COMMON-NEXT: str x19, [sp, #96] // 8-byte Folded Spill
; CHECK-COMMON-NEXT: stp s0, s1, [sp, #8] // 8-byte Folded Spill
; CHECK-COMMON-NEXT: bl __arm_sme_state
; CHECK-COMMON-NEXT: and x19, x0, #0x1
@@ -434,13 +464,14 @@ define float @frem_call_sm_compat(float %a, float %b) "aarch64_pstate_sm_compati
; CHECK-COMMON-NEXT: // %bb.3:
; CHECK-COMMON-NEXT: smstart sm
; CHECK-COMMON-NEXT: .LBB12_4:
-; CHECK-COMMON-NEXT: ldp x30, x19, [sp, #80] // 16-byte Folded Reload
; CHECK-COMMON-NEXT: ldr s0, [sp, #12] // 4-byte Folded Reload
; CHECK-COMMON-NEXT: ldp d9, d8, [sp, #64] // 16-byte Folded Reload
+; CHECK-COMMON-NEXT: ldr x19, [sp, #96] // 8-byte Folded Reload
; CHECK-COMMON-NEXT: ldp d11, d10, [sp, #48] // 16-byte Folded Reload
+; CHECK-COMMON-NEXT: ldr x30, [sp, #80] // 8-byte Folded Reload
; CHECK-COMMON-NEXT: ldp d13, d12, [sp, #32] // 16-byte Folded Reload
; CHECK-COMMON-NEXT: ldp d15, d14, [sp, #16] // 16-byte Folded Reload
-; CHECK-COMMON-NEXT: add sp, sp, #96
+; CHECK-COMMON-NEXT: add sp, sp, #112
; CHECK-COMMON-NEXT: ret
%res = frem float %a, %b
ret float %res
generated by cgit v1.1 at 2024-08-01 10:17:53 +0000