aboutsummaryrefslogtreecommitdiff
path: root/llvm/test/CodeGen/AArch64/sme-agnostic-za.ll
diff options
context:
space:
mode:
Diffstat (limited to 'llvm/test/CodeGen/AArch64/sme-agnostic-za.ll')
-rw-r--r--llvm/test/CodeGen/AArch64/sme-agnostic-za.ll47
1 files changed, 22 insertions, 25 deletions
diff --git a/llvm/test/CodeGen/AArch64/sme-agnostic-za.ll b/llvm/test/CodeGen/AArch64/sme-agnostic-za.ll
index e4f9efa..0906e10 100644
--- a/llvm/test/CodeGen/AArch64/sme-agnostic-za.ll
+++ b/llvm/test/CodeGen/AArch64/sme-agnostic-za.ll
@@ -27,7 +27,7 @@ define i64 @agnostic_caller_private_za_callee(i64 %v) nounwind "aarch64_za_state
; CHECK-LABEL: agnostic_caller_private_za_callee:
; CHECK: // %bb.0:
; CHECK-NEXT: stp x29, x30, [sp, #-32]! // 16-byte Folded Spill
-; CHECK-NEXT: str x19, [sp, #16] // 8-byte Folded Spill
+; CHECK-NEXT: str x19, [sp, #16] // 8-byte Spill
; CHECK-NEXT: mov x29, sp
; CHECK-NEXT: mov x8, x0
; CHECK-NEXT: bl __arm_sme_state_size
@@ -49,14 +49,14 @@ define i64 @agnostic_caller_private_za_callee(i64 %v) nounwind "aarch64_za_state
; CHECK-NEXT: bl __arm_sme_restore
; CHECK-NEXT: mov x0, x1
; CHECK-NEXT: mov sp, x29
-; CHECK-NEXT: ldr x19, [sp, #16] // 8-byte Folded Reload
+; CHECK-NEXT: ldr x19, [sp, #16] // 8-byte Reload
; CHECK-NEXT: ldp x29, x30, [sp], #32 // 16-byte Folded Reload
; CHECK-NEXT: ret
;
; CHECK-NEWLOWERING-LABEL: agnostic_caller_private_za_callee:
; CHECK-NEWLOWERING: // %bb.0:
; CHECK-NEWLOWERING-NEXT: stp x29, x30, [sp, #-32]! // 16-byte Folded Spill
-; CHECK-NEWLOWERING-NEXT: str x19, [sp, #16] // 8-byte Folded Spill
+; CHECK-NEWLOWERING-NEXT: str x19, [sp, #16] // 8-byte Spill
; CHECK-NEWLOWERING-NEXT: mov x29, sp
; CHECK-NEWLOWERING-NEXT: mov x8, x0
; CHECK-NEWLOWERING-NEXT: bl __arm_sme_state_size
@@ -67,12 +67,12 @@ define i64 @agnostic_caller_private_za_callee(i64 %v) nounwind "aarch64_za_state
; CHECK-NEWLOWERING-NEXT: mov x0, x8
; CHECK-NEWLOWERING-NEXT: bl private_za_decl
; CHECK-NEWLOWERING-NEXT: bl private_za_decl
-; CHECK-NEWLOWERING-NEXT: mov x8, x0
+; CHECK-NEWLOWERING-NEXT: mov x1, x0
; CHECK-NEWLOWERING-NEXT: mov x0, x19
; CHECK-NEWLOWERING-NEXT: bl __arm_sme_restore
-; CHECK-NEWLOWERING-NEXT: mov x0, x8
+; CHECK-NEWLOWERING-NEXT: mov x0, x1
; CHECK-NEWLOWERING-NEXT: mov sp, x29
-; CHECK-NEWLOWERING-NEXT: ldr x19, [sp, #16] // 8-byte Folded Reload
+; CHECK-NEWLOWERING-NEXT: ldr x19, [sp, #16] // 8-byte Reload
; CHECK-NEWLOWERING-NEXT: ldp x29, x30, [sp], #32 // 16-byte Folded Reload
; CHECK-NEWLOWERING-NEXT: ret
%res = call i64 @private_za_decl(i64 %v)
@@ -170,11 +170,11 @@ define i64 @streaming_agnostic_caller_nonstreaming_private_za_callee(i64 %v) nou
; CHECK-NEWLOWERING-NEXT: mov x0, x8
; CHECK-NEWLOWERING-NEXT: bl private_za_decl
; CHECK-NEWLOWERING-NEXT: bl private_za_decl
+; CHECK-NEWLOWERING-NEXT: mov x1, x0
; CHECK-NEWLOWERING-NEXT: smstart sm
-; CHECK-NEWLOWERING-NEXT: mov x8, x0
; CHECK-NEWLOWERING-NEXT: mov x0, x20
; CHECK-NEWLOWERING-NEXT: bl __arm_sme_restore
-; CHECK-NEWLOWERING-NEXT: mov x0, x8
+; CHECK-NEWLOWERING-NEXT: mov x0, x1
; CHECK-NEWLOWERING-NEXT: sub sp, x29, #64
; CHECK-NEWLOWERING-NEXT: ldp x20, x19, [sp, #80] // 16-byte Folded Reload
; CHECK-NEWLOWERING-NEXT: ldp x29, x30, [sp, #64] // 16-byte Folded Reload
@@ -267,14 +267,14 @@ define i64 @streaming_compatible_agnostic_caller_nonstreaming_private_za_callee(
; CHECK-NEWLOWERING-NEXT: mov x0, x8
; CHECK-NEWLOWERING-NEXT: bl private_za_decl
; CHECK-NEWLOWERING-NEXT: bl private_za_decl
+; CHECK-NEWLOWERING-NEXT: mov x1, x0
; CHECK-NEWLOWERING-NEXT: tbz w20, #0, .LBB5_4
; CHECK-NEWLOWERING-NEXT: // %bb.3:
; CHECK-NEWLOWERING-NEXT: smstart sm
; CHECK-NEWLOWERING-NEXT: .LBB5_4:
-; CHECK-NEWLOWERING-NEXT: mov x8, x0
; CHECK-NEWLOWERING-NEXT: mov x0, x19
; CHECK-NEWLOWERING-NEXT: bl __arm_sme_restore
-; CHECK-NEWLOWERING-NEXT: mov x0, x8
+; CHECK-NEWLOWERING-NEXT: mov x0, x1
; CHECK-NEWLOWERING-NEXT: sub sp, x29, #64
; CHECK-NEWLOWERING-NEXT: ldp x20, x19, [sp, #80] // 16-byte Folded Reload
; CHECK-NEWLOWERING-NEXT: ldp x29, x30, [sp, #64] // 16-byte Folded Reload
@@ -298,7 +298,7 @@ define i64 @test_many_callee_arguments(
; CHECK-LABEL: test_many_callee_arguments:
; CHECK: // %bb.0:
; CHECK-NEXT: stp x29, x30, [sp, #-32]! // 16-byte Folded Spill
-; CHECK-NEXT: str x19, [sp, #16] // 8-byte Folded Spill
+; CHECK-NEXT: str x19, [sp, #16] // 8-byte Spill
; CHECK-NEXT: mov x29, sp
; CHECK-NEXT: mov x8, x0
; CHECK-NEXT: bl __arm_sme_state_size
@@ -316,14 +316,14 @@ define i64 @test_many_callee_arguments(
; CHECK-NEXT: bl __arm_sme_restore
; CHECK-NEXT: mov x0, x1
; CHECK-NEXT: mov sp, x29
-; CHECK-NEXT: ldr x19, [sp, #16] // 8-byte Folded Reload
+; CHECK-NEXT: ldr x19, [sp, #16] // 8-byte Reload
; CHECK-NEXT: ldp x29, x30, [sp], #32 // 16-byte Folded Reload
; CHECK-NEXT: ret
;
; CHECK-NEWLOWERING-LABEL: test_many_callee_arguments:
; CHECK-NEWLOWERING: // %bb.0:
; CHECK-NEWLOWERING-NEXT: stp x29, x30, [sp, #-32]! // 16-byte Folded Spill
-; CHECK-NEWLOWERING-NEXT: str x19, [sp, #16] // 8-byte Folded Spill
+; CHECK-NEWLOWERING-NEXT: str x19, [sp, #16] // 8-byte Spill
; CHECK-NEWLOWERING-NEXT: mov x29, sp
; CHECK-NEWLOWERING-NEXT: mov x8, x0
; CHECK-NEWLOWERING-NEXT: bl __arm_sme_state_size
@@ -336,12 +336,12 @@ define i64 @test_many_callee_arguments(
; CHECK-NEWLOWERING-NEXT: mov x0, x8
; CHECK-NEWLOWERING-NEXT: bl many_args_private_za_callee
; CHECK-NEWLOWERING-NEXT: add sp, sp, #16
-; CHECK-NEWLOWERING-NEXT: mov x8, x0
+; CHECK-NEWLOWERING-NEXT: mov x1, x0
; CHECK-NEWLOWERING-NEXT: mov x0, x19
; CHECK-NEWLOWERING-NEXT: bl __arm_sme_restore
-; CHECK-NEWLOWERING-NEXT: mov x0, x8
+; CHECK-NEWLOWERING-NEXT: mov x0, x1
; CHECK-NEWLOWERING-NEXT: mov sp, x29
-; CHECK-NEWLOWERING-NEXT: ldr x19, [sp, #16] // 8-byte Folded Reload
+; CHECK-NEWLOWERING-NEXT: ldr x19, [sp, #16] // 8-byte Reload
; CHECK-NEWLOWERING-NEXT: ldp x29, x30, [sp], #32 // 16-byte Folded Reload
; CHECK-NEWLOWERING-NEXT: ret
i64 %0, i64 %1, i64 %2, i64 %3, i64 %4, i64 %5, i64 %6, i64 %7, i64 %8, i64 %9
@@ -351,12 +351,11 @@ define i64 @test_many_callee_arguments(
ret i64 %ret
}
-; FIXME: The new lowering should avoid saves/restores in the probing loop.
define void @agnostic_za_buffer_alloc_with_stack_probes() nounwind "aarch64_za_state_agnostic" "probe-stack"="inline-asm" "stack-probe-size"="65536"{
; CHECK-LABEL: agnostic_za_buffer_alloc_with_stack_probes:
; CHECK: // %bb.0:
; CHECK-NEXT: stp x29, x30, [sp, #-32]! // 16-byte Folded Spill
-; CHECK-NEXT: str x19, [sp, #16] // 8-byte Folded Spill
+; CHECK-NEXT: str x19, [sp, #16] // 8-byte Spill
; CHECK-NEXT: mov x29, sp
; CHECK-NEXT: bl __arm_sme_state_size
; CHECK-NEXT: mov x8, sp
@@ -377,28 +376,26 @@ define void @agnostic_za_buffer_alloc_with_stack_probes() nounwind "aarch64_za_s
; CHECK-NEXT: mov x0, x19
; CHECK-NEXT: bl __arm_sme_restore
; CHECK-NEXT: mov sp, x29
-; CHECK-NEXT: ldr x19, [sp, #16] // 8-byte Folded Reload
+; CHECK-NEXT: ldr x19, [sp, #16] // 8-byte Reload
; CHECK-NEXT: ldp x29, x30, [sp], #32 // 16-byte Folded Reload
; CHECK-NEXT: ret
;
; CHECK-NEWLOWERING-LABEL: agnostic_za_buffer_alloc_with_stack_probes:
; CHECK-NEWLOWERING: // %bb.0:
; CHECK-NEWLOWERING-NEXT: stp x29, x30, [sp, #-32]! // 16-byte Folded Spill
-; CHECK-NEWLOWERING-NEXT: str x19, [sp, #16] // 8-byte Folded Spill
+; CHECK-NEWLOWERING-NEXT: str x19, [sp, #16] // 8-byte Spill
; CHECK-NEWLOWERING-NEXT: mov x29, sp
; CHECK-NEWLOWERING-NEXT: bl __arm_sme_state_size
; CHECK-NEWLOWERING-NEXT: mov x8, sp
; CHECK-NEWLOWERING-NEXT: sub x19, x8, x0
-; CHECK-NEWLOWERING-NEXT: .LBB7_1: // =>This Inner Loop Header: Depth=1
-; CHECK-NEWLOWERING-NEXT: sub sp, sp, #16, lsl #12 // =65536
; CHECK-NEWLOWERING-NEXT: mov x0, x19
; CHECK-NEWLOWERING-NEXT: bl __arm_sme_save
+; CHECK-NEWLOWERING-NEXT: .LBB7_1: // =>This Inner Loop Header: Depth=1
+; CHECK-NEWLOWERING-NEXT: sub sp, sp, #16, lsl #12 // =65536
; CHECK-NEWLOWERING-NEXT: cmp sp, x19
; CHECK-NEWLOWERING-NEXT: b.le .LBB7_3
; CHECK-NEWLOWERING-NEXT: // %bb.2: // in Loop: Header=BB7_1 Depth=1
-; CHECK-NEWLOWERING-NEXT: mov x0, x19
; CHECK-NEWLOWERING-NEXT: str xzr, [sp]
-; CHECK-NEWLOWERING-NEXT: bl __arm_sme_restore
; CHECK-NEWLOWERING-NEXT: b .LBB7_1
; CHECK-NEWLOWERING-NEXT: .LBB7_3:
; CHECK-NEWLOWERING-NEXT: mov sp, x19
@@ -407,7 +404,7 @@ define void @agnostic_za_buffer_alloc_with_stack_probes() nounwind "aarch64_za_s
; CHECK-NEWLOWERING-NEXT: mov x0, x19
; CHECK-NEWLOWERING-NEXT: bl __arm_sme_restore
; CHECK-NEWLOWERING-NEXT: mov sp, x29
-; CHECK-NEWLOWERING-NEXT: ldr x19, [sp, #16] // 8-byte Folded Reload
+; CHECK-NEWLOWERING-NEXT: ldr x19, [sp, #16] // 8-byte Reload
; CHECK-NEWLOWERING-NEXT: ldp x29, x30, [sp], #32 // 16-byte Folded Reload
; CHECK-NEWLOWERING-NEXT: ret
call void @private_za()
generated by cgit v1.1 at 2026-01-10 19:43:55 +0000