derek.holloway
7502bf64a4
CodeQL / Analyze (actions) (push) Canceled after 0s
CodeQL / Analyze (python) (push) Canceled after 0s
pre-commit / pre-commit (push) Canceled after 0s
Multi-Arch CI / setup (push) Canceled after 0s
Publish build_manylinux_x86_64 images / publish_build_manylinux_x86_64 (push) Canceled after 0s
Setup (Multi-Arch) / setup (push) Canceled after 0s
Multi-Arch CI / Linux::${{ fromJSON(needs.setup.outputs.linux_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubi10 image / publish_no_rocm_image_ubi10 (push) Canceled after 0s
Multi-Arch CI / Windows::${{ fromJSON(needs.setup.outputs.windows_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Multi-Arch CI / Manifest Diff (push) Canceled after 0s
TheRock Manifest Diff Report / Generate Manifest Diff Report (push) Canceled after 0s
Multi-Arch CI / CI Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_ocl_rt images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_rocgdb images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Unit Tests / Unit Tests :: ubuntu-24.04 (push) Canceled after 0s
Unit Tests / Unit Tests :: windows-2022 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_media (push) Canceled after 0s
Unit Tests / Unit Tests Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_openmpi (push) Canceled after 0s
197 lines
7.7 KiB
LLVM
197 lines
7.7 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
|
|
; RUN: llc -mtriple=loongarch32 < %s | FileCheck %s --check-prefix=LA32
|
|
; RUN: llc -mtriple=loongarch64 < %s | FileCheck %s --check-prefix=LA64
|
|
|
|
%va_list = type { ptr, ptr, ptr, i32, i32 }
|
|
|
|
define preserve_nonecc i32 @callee(i32 %a1, i32 %a2, i32 %a3, i32 %a4, i32 %a5, ...) nounwind noinline ssp {
|
|
; LA32-LABEL: callee:
|
|
; LA32: # %bb.0: # %entry
|
|
; LA32-NEXT: addi.w $sp, $sp, -48
|
|
; LA32-NEXT: st.w $a7, $sp, 44
|
|
; LA32-NEXT: st.w $a6, $sp, 40
|
|
; LA32-NEXT: st.w $a5, $sp, 36
|
|
; LA32-NEXT: ld.w $a0, $sp, 20
|
|
; LA32-NEXT: addi.w $a1, $sp, 36
|
|
; LA32-NEXT: st.w $a1, $sp, 8
|
|
; LA32-NEXT: bgez $a0, .LBB0_3
|
|
; LA32-NEXT: # %bb.1: # %maybe_reg
|
|
; LA32-NEXT: addi.w $a1, $a0, 8
|
|
; LA32-NEXT: st.w $a1, $sp, 20
|
|
; LA32-NEXT: bgtz $a1, .LBB0_3
|
|
; LA32-NEXT: # %bb.2: # %in_reg
|
|
; LA32-NEXT: ld.w $a1, $sp, 12
|
|
; LA32-NEXT: add.w $a0, $a1, $a0
|
|
; LA32-NEXT: b .LBB0_4
|
|
; LA32-NEXT: .LBB0_3: # %on_stack
|
|
; LA32-NEXT: ld.w $a0, $sp, 8
|
|
; LA32-NEXT: addi.w $a1, $a0, 8
|
|
; LA32-NEXT: st.w $a1, $sp, 8
|
|
; LA32-NEXT: .LBB0_4: # %end
|
|
; LA32-NEXT: ld.w $a0, $a0, 0
|
|
; LA32-NEXT: addi.w $sp, $sp, 48
|
|
; LA32-NEXT: ret
|
|
;
|
|
; LA64-LABEL: callee:
|
|
; LA64: # %bb.0: # %entry
|
|
; LA64-NEXT: addi.d $sp, $sp, -80
|
|
; LA64-NEXT: st.d $a7, $sp, 72
|
|
; LA64-NEXT: st.d $a6, $sp, 64
|
|
; LA64-NEXT: st.d $a5, $sp, 56
|
|
; LA64-NEXT: ld.w $a0, $sp, 40
|
|
; LA64-NEXT: addi.d $a1, $sp, 56
|
|
; LA64-NEXT: st.d $a1, $sp, 16
|
|
; LA64-NEXT: bgez $a0, .LBB0_3
|
|
; LA64-NEXT: # %bb.1: # %maybe_reg
|
|
; LA64-NEXT: addi.w $a1, $a0, 8
|
|
; LA64-NEXT: st.w $a1, $sp, 40
|
|
; LA64-NEXT: bgtz $a1, .LBB0_3
|
|
; LA64-NEXT: # %bb.2: # %in_reg
|
|
; LA64-NEXT: ld.d $a1, $sp, 24
|
|
; LA64-NEXT: add.d $a0, $a1, $a0
|
|
; LA64-NEXT: b .LBB0_4
|
|
; LA64-NEXT: .LBB0_3: # %on_stack
|
|
; LA64-NEXT: ld.d $a0, $sp, 16
|
|
; LA64-NEXT: addi.d $a1, $a0, 8
|
|
; LA64-NEXT: st.d $a1, $sp, 16
|
|
; LA64-NEXT: .LBB0_4: # %end
|
|
; LA64-NEXT: ld.w $a0, $a0, 0
|
|
; LA64-NEXT: addi.d $sp, $sp, 80
|
|
; LA64-NEXT: ret
|
|
entry:
|
|
%args = alloca %va_list, align 8
|
|
call void @llvm.va_start(ptr %args)
|
|
%gr_offs_p = getelementptr inbounds %va_list, ptr %args, i32 0, i32 3
|
|
%gr_offs = load i32, ptr %gr_offs_p, align 8
|
|
%0 = icmp sge i32 %gr_offs, 0
|
|
br i1 %0, label %on_stack, label %maybe_reg
|
|
|
|
maybe_reg:
|
|
%new_reg_offs = add i32 %gr_offs, 8
|
|
store i32 %new_reg_offs, ptr %gr_offs_p, align 8
|
|
%inreg = icmp sle i32 %new_reg_offs, 0
|
|
br i1 %inreg, label %in_reg, label %on_stack
|
|
|
|
in_reg:
|
|
%reg_top_p = getelementptr inbounds %va_list, ptr %args, i32 0, i32 1
|
|
%reg_top = load ptr, ptr %reg_top_p, align 8
|
|
%reg = getelementptr inbounds i8, ptr %reg_top, i32 %gr_offs
|
|
br label %end
|
|
|
|
on_stack:
|
|
%stack_p = getelementptr inbounds %va_list, ptr %args, i32 0, i32 0
|
|
%stack = load ptr, ptr %stack_p, align 8
|
|
%new_stack = getelementptr inbounds i8, ptr %stack, i64 8
|
|
store ptr %new_stack, ptr %stack_p, align 8
|
|
br label %end
|
|
|
|
end:
|
|
%p = phi ptr [ %reg, %in_reg ], [ %stack, %on_stack ]
|
|
%10 = load i32, ptr %p, align 8
|
|
call void @llvm.va_end.p0(ptr %args)
|
|
ret i32 %10
|
|
}
|
|
|
|
declare void @llvm.va_start(ptr) nounwind
|
|
declare void @llvm.va_end(ptr) nounwind
|
|
|
|
define i32 @caller() nounwind ssp {
|
|
; LA32-LABEL: caller:
|
|
; LA32: # %bb.0:
|
|
; LA32-NEXT: addi.w $sp, $sp, -48
|
|
; LA32-NEXT: st.w $ra, $sp, 44 # 4-byte Folded Spill
|
|
; LA32-NEXT: st.w $s0, $sp, 40 # 4-byte Folded Spill
|
|
; LA32-NEXT: st.w $s1, $sp, 36 # 4-byte Folded Spill
|
|
; LA32-NEXT: st.w $s2, $sp, 32 # 4-byte Folded Spill
|
|
; LA32-NEXT: st.w $s3, $sp, 28 # 4-byte Folded Spill
|
|
; LA32-NEXT: st.w $s4, $sp, 24 # 4-byte Folded Spill
|
|
; LA32-NEXT: st.w $s5, $sp, 20 # 4-byte Folded Spill
|
|
; LA32-NEXT: st.w $s6, $sp, 16 # 4-byte Folded Spill
|
|
; LA32-NEXT: st.w $s7, $sp, 12 # 4-byte Folded Spill
|
|
; LA32-NEXT: st.w $s8, $sp, 8 # 4-byte Folded Spill
|
|
; LA32-NEXT: ori $a0, $zero, 10
|
|
; LA32-NEXT: st.w $a0, $sp, 4
|
|
; LA32-NEXT: ori $t0, $zero, 9
|
|
; LA32-NEXT: ori $a0, $zero, 1
|
|
; LA32-NEXT: ori $a1, $zero, 2
|
|
; LA32-NEXT: ori $a2, $zero, 3
|
|
; LA32-NEXT: ori $a3, $zero, 4
|
|
; LA32-NEXT: ori $a4, $zero, 5
|
|
; LA32-NEXT: ori $a5, $zero, 6
|
|
; LA32-NEXT: ori $a6, $zero, 7
|
|
; LA32-NEXT: ori $a7, $zero, 8
|
|
; LA32-NEXT: st.w $t0, $sp, 0
|
|
; LA32-NEXT: bl callee
|
|
; LA32-NEXT: ld.w $s8, $sp, 8 # 4-byte Folded Reload
|
|
; LA32-NEXT: ld.w $s7, $sp, 12 # 4-byte Folded Reload
|
|
; LA32-NEXT: ld.w $s6, $sp, 16 # 4-byte Folded Reload
|
|
; LA32-NEXT: ld.w $s5, $sp, 20 # 4-byte Folded Reload
|
|
; LA32-NEXT: ld.w $s4, $sp, 24 # 4-byte Folded Reload
|
|
; LA32-NEXT: ld.w $s3, $sp, 28 # 4-byte Folded Reload
|
|
; LA32-NEXT: ld.w $s2, $sp, 32 # 4-byte Folded Reload
|
|
; LA32-NEXT: ld.w $s1, $sp, 36 # 4-byte Folded Reload
|
|
; LA32-NEXT: ld.w $s0, $sp, 40 # 4-byte Folded Reload
|
|
; LA32-NEXT: ld.w $ra, $sp, 44 # 4-byte Folded Reload
|
|
; LA32-NEXT: addi.w $sp, $sp, 48
|
|
; LA32-NEXT: ret
|
|
;
|
|
; LA64-LABEL: caller:
|
|
; LA64: # %bb.0:
|
|
; LA64-NEXT: addi.d $sp, $sp, -176
|
|
; LA64-NEXT: st.d $ra, $sp, 168 # 8-byte Folded Spill
|
|
; LA64-NEXT: st.d $s0, $sp, 160 # 8-byte Folded Spill
|
|
; LA64-NEXT: st.d $s1, $sp, 152 # 8-byte Folded Spill
|
|
; LA64-NEXT: st.d $s2, $sp, 144 # 8-byte Folded Spill
|
|
; LA64-NEXT: st.d $s3, $sp, 136 # 8-byte Folded Spill
|
|
; LA64-NEXT: st.d $s4, $sp, 128 # 8-byte Folded Spill
|
|
; LA64-NEXT: st.d $s5, $sp, 120 # 8-byte Folded Spill
|
|
; LA64-NEXT: st.d $s6, $sp, 112 # 8-byte Folded Spill
|
|
; LA64-NEXT: st.d $s7, $sp, 104 # 8-byte Folded Spill
|
|
; LA64-NEXT: st.d $s8, $sp, 96 # 8-byte Folded Spill
|
|
; LA64-NEXT: fst.d $fs0, $sp, 88 # 8-byte Folded Spill
|
|
; LA64-NEXT: fst.d $fs1, $sp, 80 # 8-byte Folded Spill
|
|
; LA64-NEXT: fst.d $fs2, $sp, 72 # 8-byte Folded Spill
|
|
; LA64-NEXT: fst.d $fs3, $sp, 64 # 8-byte Folded Spill
|
|
; LA64-NEXT: fst.d $fs4, $sp, 56 # 8-byte Folded Spill
|
|
; LA64-NEXT: fst.d $fs5, $sp, 48 # 8-byte Folded Spill
|
|
; LA64-NEXT: fst.d $fs6, $sp, 40 # 8-byte Folded Spill
|
|
; LA64-NEXT: fst.d $fs7, $sp, 32 # 8-byte Folded Spill
|
|
; LA64-NEXT: ori $a0, $zero, 10
|
|
; LA64-NEXT: st.d $a0, $sp, 8
|
|
; LA64-NEXT: ori $t0, $zero, 9
|
|
; LA64-NEXT: ori $a0, $zero, 1
|
|
; LA64-NEXT: ori $a1, $zero, 2
|
|
; LA64-NEXT: ori $a2, $zero, 3
|
|
; LA64-NEXT: ori $a3, $zero, 4
|
|
; LA64-NEXT: ori $a4, $zero, 5
|
|
; LA64-NEXT: ori $a5, $zero, 6
|
|
; LA64-NEXT: ori $a6, $zero, 7
|
|
; LA64-NEXT: ori $a7, $zero, 8
|
|
; LA64-NEXT: st.d $t0, $sp, 0
|
|
; LA64-NEXT: pcaddu18i $ra, %call36(callee)
|
|
; LA64-NEXT: jirl $ra, $ra, 0
|
|
; LA64-NEXT: fld.d $fs7, $sp, 32 # 8-byte Folded Reload
|
|
; LA64-NEXT: fld.d $fs6, $sp, 40 # 8-byte Folded Reload
|
|
; LA64-NEXT: fld.d $fs5, $sp, 48 # 8-byte Folded Reload
|
|
; LA64-NEXT: fld.d $fs4, $sp, 56 # 8-byte Folded Reload
|
|
; LA64-NEXT: fld.d $fs3, $sp, 64 # 8-byte Folded Reload
|
|
; LA64-NEXT: fld.d $fs2, $sp, 72 # 8-byte Folded Reload
|
|
; LA64-NEXT: fld.d $fs1, $sp, 80 # 8-byte Folded Reload
|
|
; LA64-NEXT: fld.d $fs0, $sp, 88 # 8-byte Folded Reload
|
|
; LA64-NEXT: ld.d $s8, $sp, 96 # 8-byte Folded Reload
|
|
; LA64-NEXT: ld.d $s7, $sp, 104 # 8-byte Folded Reload
|
|
; LA64-NEXT: ld.d $s6, $sp, 112 # 8-byte Folded Reload
|
|
; LA64-NEXT: ld.d $s5, $sp, 120 # 8-byte Folded Reload
|
|
; LA64-NEXT: ld.d $s4, $sp, 128 # 8-byte Folded Reload
|
|
; LA64-NEXT: ld.d $s3, $sp, 136 # 8-byte Folded Reload
|
|
; LA64-NEXT: ld.d $s2, $sp, 144 # 8-byte Folded Reload
|
|
; LA64-NEXT: ld.d $s1, $sp, 152 # 8-byte Folded Reload
|
|
; LA64-NEXT: ld.d $s0, $sp, 160 # 8-byte Folded Reload
|
|
; LA64-NEXT: ld.d $ra, $sp, 168 # 8-byte Folded Reload
|
|
; LA64-NEXT: addi.d $sp, $sp, 176
|
|
; LA64-NEXT: ret
|
|
%r = tail call preserve_nonecc i32 (i32, i32, i32, i32, i32, ...) @callee(i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10)
|
|
ret i32 %r
|
|
}
|
|
|