[LoongArch] Allow f16 codegen with expansion to libcalls #94456

xen0n · 2024-06-05T10:36:25Z

The test case is adapted from llvm/test/CodeGen/RISCV/fp16-promote.ll, because it covers some more IR patterns that ought to be common.

Fixes #93894

The test case is adapted from llvm/test/CodeGen/RISCV/fp16-promote.ll, because it covers some more IR patterns that ought to be common. Fixes llvm#93894

llvmbot · 2024-06-05T10:36:57Z

@llvm/pr-subscribers-backend-loongarch

Author: WÁNG Xuěruì (xen0n)

Changes

The test case is adapted from llvm/test/CodeGen/RISCV/fp16-promote.ll, because it covers some more IR patterns that ought to be common.

Fixes #93894

Full diff: https://github.com/llvm/llvm-project/pull/94456.diff

2 Files Affected:

(modified) llvm/lib/Target/LoongArch/LoongArchISelLowering.cpp (+8)
(added) llvm/test/CodeGen/LoongArch/fp16-promote.ll (+326)

diff --git a/llvm/lib/Target/LoongArch/LoongArchISelLowering.cpp b/llvm/lib/Target/LoongArch/LoongArchISelLowering.cpp
index 51384f25d2455..7ab6640550b62 100644
--- a/llvm/lib/Target/LoongArch/LoongArchISelLowering.cpp
+++ b/llvm/lib/Target/LoongArch/LoongArchISelLowering.cpp
@@ -171,6 +171,8 @@ LoongArchTargetLowering::LoongArchTargetLowering(const TargetMachine &TM,
   // Set operations for 'F' feature.
 
   if (Subtarget.hasBasicF()) {
+    setLoadExtAction(ISD::EXTLOAD, MVT::f32, MVT::f16, Expand);
+    setTruncStoreAction(MVT::f32, MVT::f16, Expand);
     setCondCodeAction(FPCCToExpand, MVT::f32, Expand);
 
     setOperationAction(ISD::SELECT_CC, MVT::f32, Expand);
@@ -186,6 +188,8 @@ LoongArchTargetLowering::LoongArchTargetLowering(const TargetMachine &TM,
     setOperationAction(ISD::FSINCOS, MVT::f32, Expand);
     setOperationAction(ISD::FPOW, MVT::f32, Expand);
     setOperationAction(ISD::FREM, MVT::f32, Expand);
+    setOperationAction(ISD::FP16_TO_FP, MVT::f32, Expand);
+    setOperationAction(ISD::FP_TO_FP16, MVT::f32, Expand);
 
     if (Subtarget.is64Bit())
       setOperationAction(ISD::FRINT, MVT::f32, Legal);
@@ -202,7 +206,9 @@ LoongArchTargetLowering::LoongArchTargetLowering(const TargetMachine &TM,
   // Set operations for 'D' feature.
 
   if (Subtarget.hasBasicD()) {
+    setLoadExtAction(ISD::EXTLOAD, MVT::f64, MVT::f16, Expand);
     setLoadExtAction(ISD::EXTLOAD, MVT::f64, MVT::f32, Expand);
+    setTruncStoreAction(MVT::f64, MVT::f16, Expand);
     setTruncStoreAction(MVT::f64, MVT::f32, Expand);
     setCondCodeAction(FPCCToExpand, MVT::f64, Expand);
 
@@ -219,6 +225,8 @@ LoongArchTargetLowering::LoongArchTargetLowering(const TargetMachine &TM,
     setOperationAction(ISD::FSINCOS, MVT::f64, Expand);
     setOperationAction(ISD::FPOW, MVT::f64, Expand);
     setOperationAction(ISD::FREM, MVT::f64, Expand);
+    setOperationAction(ISD::FP16_TO_FP, MVT::f64, Expand);
+    setOperationAction(ISD::FP_TO_FP16, MVT::f64, Expand);
 
     if (Subtarget.is64Bit())
       setOperationAction(ISD::FRINT, MVT::f64, Legal);
diff --git a/llvm/test/CodeGen/LoongArch/fp16-promote.ll b/llvm/test/CodeGen/LoongArch/fp16-promote.ll
new file mode 100644
index 0000000000000..75f920b43a06c
--- /dev/null
+++ b/llvm/test/CodeGen/LoongArch/fp16-promote.ll
@@ -0,0 +1,326 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
+; RUN: llc --mtriple=loongarch32 --mattr=+d < %s | FileCheck %s --check-prefix=LA32
+; RUN: llc --mtriple=loongarch64 --mattr=+d < %s | FileCheck %s --check-prefix=LA64
+
+define void @test_load_store(ptr %p, ptr %q) nounwind {
+; LA32-LABEL: test_load_store:
+; LA32:       # %bb.0:
+; LA32-NEXT:    ld.h $a0, $a0, 0
+; LA32-NEXT:    st.h $a0, $a1, 0
+; LA32-NEXT:    ret
+;
+; LA64-LABEL: test_load_store:
+; LA64:       # %bb.0:
+; LA64-NEXT:    ld.h $a0, $a0, 0
+; LA64-NEXT:    st.h $a0, $a1, 0
+; LA64-NEXT:    ret
+  %a = load half, ptr %p
+  store half %a, ptr %q
+  ret void
+}
+
+define float @test_fpextend_float(ptr %p) nounwind {
+; LA32-LABEL: test_fpextend_float:
+; LA32:       # %bb.0:
+; LA32-NEXT:    ld.hu $a0, $a0, 0
+; LA32-NEXT:    b %plt(__gnu_h2f_ieee)
+;
+; LA64-LABEL: test_fpextend_float:
+; LA64:       # %bb.0:
+; LA64-NEXT:    ld.hu $a0, $a0, 0
+; LA64-NEXT:    b %plt(__gnu_h2f_ieee)
+  %a = load half, ptr %p
+  %r = fpext half %a to float
+  ret float %r
+}
+
+define double @test_fpextend_double(ptr %p) nounwind {
+; LA32-LABEL: test_fpextend_double:
+; LA32:       # %bb.0:
+; LA32-NEXT:    addi.w $sp, $sp, -16
+; LA32-NEXT:    st.w $ra, $sp, 12 # 4-byte Folded Spill
+; LA32-NEXT:    ld.hu $a0, $a0, 0
+; LA32-NEXT:    bl %plt(__gnu_h2f_ieee)
+; LA32-NEXT:    fcvt.d.s $fa0, $fa0
+; LA32-NEXT:    ld.w $ra, $sp, 12 # 4-byte Folded Reload
+; LA32-NEXT:    addi.w $sp, $sp, 16
+; LA32-NEXT:    ret
+;
+; LA64-LABEL: test_fpextend_double:
+; LA64:       # %bb.0:
+; LA64-NEXT:    addi.d $sp, $sp, -16
+; LA64-NEXT:    st.d $ra, $sp, 8 # 8-byte Folded Spill
+; LA64-NEXT:    ld.hu $a0, $a0, 0
+; LA64-NEXT:    bl %plt(__gnu_h2f_ieee)
+; LA64-NEXT:    fcvt.d.s $fa0, $fa0
+; LA64-NEXT:    ld.d $ra, $sp, 8 # 8-byte Folded Reload
+; LA64-NEXT:    addi.d $sp, $sp, 16
+; LA64-NEXT:    ret
+  %a = load half, ptr %p
+  %r = fpext half %a to double
+  ret double %r
+}
+
+define void @test_fptrunc_float(float %f, ptr %p) nounwind {
+; LA32-LABEL: test_fptrunc_float:
+; LA32:       # %bb.0:
+; LA32-NEXT:    addi.w $sp, $sp, -16
+; LA32-NEXT:    st.w $ra, $sp, 12 # 4-byte Folded Spill
+; LA32-NEXT:    st.w $fp, $sp, 8 # 4-byte Folded Spill
+; LA32-NEXT:    move $fp, $a0
+; LA32-NEXT:    bl %plt(__gnu_f2h_ieee)
+; LA32-NEXT:    st.h $a0, $fp, 0
+; LA32-NEXT:    ld.w $fp, $sp, 8 # 4-byte Folded Reload
+; LA32-NEXT:    ld.w $ra, $sp, 12 # 4-byte Folded Reload
+; LA32-NEXT:    addi.w $sp, $sp, 16
+; LA32-NEXT:    ret
+;
+; LA64-LABEL: test_fptrunc_float:
+; LA64:       # %bb.0:
+; LA64-NEXT:    addi.d $sp, $sp, -16
+; LA64-NEXT:    st.d $ra, $sp, 8 # 8-byte Folded Spill
+; LA64-NEXT:    st.d $fp, $sp, 0 # 8-byte Folded Spill
+; LA64-NEXT:    move $fp, $a0
+; LA64-NEXT:    bl %plt(__gnu_f2h_ieee)
+; LA64-NEXT:    st.h $a0, $fp, 0
+; LA64-NEXT:    ld.d $fp, $sp, 0 # 8-byte Folded Reload
+; LA64-NEXT:    ld.d $ra, $sp, 8 # 8-byte Folded Reload
+; LA64-NEXT:    addi.d $sp, $sp, 16
+; LA64-NEXT:    ret
+  %a = fptrunc float %f to half
+  store half %a, ptr %p
+  ret void
+}
+
+define void @test_fptrunc_double(double %d, ptr %p) nounwind {
+; LA32-LABEL: test_fptrunc_double:
+; LA32:       # %bb.0:
+; LA32-NEXT:    addi.w $sp, $sp, -16
+; LA32-NEXT:    st.w $ra, $sp, 12 # 4-byte Folded Spill
+; LA32-NEXT:    st.w $fp, $sp, 8 # 4-byte Folded Spill
+; LA32-NEXT:    move $fp, $a0
+; LA32-NEXT:    bl %plt(__truncdfhf2)
+; LA32-NEXT:    st.h $a0, $fp, 0
+; LA32-NEXT:    ld.w $fp, $sp, 8 # 4-byte Folded Reload
+; LA32-NEXT:    ld.w $ra, $sp, 12 # 4-byte Folded Reload
+; LA32-NEXT:    addi.w $sp, $sp, 16
+; LA32-NEXT:    ret
+;
+; LA64-LABEL: test_fptrunc_double:
+; LA64:       # %bb.0:
+; LA64-NEXT:    addi.d $sp, $sp, -16
+; LA64-NEXT:    st.d $ra, $sp, 8 # 8-byte Folded Spill
+; LA64-NEXT:    st.d $fp, $sp, 0 # 8-byte Folded Spill
+; LA64-NEXT:    move $fp, $a0
+; LA64-NEXT:    bl %plt(__truncdfhf2)
+; LA64-NEXT:    st.h $a0, $fp, 0
+; LA64-NEXT:    ld.d $fp, $sp, 0 # 8-byte Folded Reload
+; LA64-NEXT:    ld.d $ra, $sp, 8 # 8-byte Folded Reload
+; LA64-NEXT:    addi.d $sp, $sp, 16
+; LA64-NEXT:    ret
+  %a = fptrunc double %d to half
+  store half %a, ptr %p
+  ret void
+}
+
+define half @test_fadd_reg(half %a, half %b) nounwind {
+; LA32-LABEL: test_fadd_reg:
+; LA32:       # %bb.0:
+; LA32-NEXT:    addi.w $sp, $sp, -32
+; LA32-NEXT:    st.w $ra, $sp, 28 # 4-byte Folded Spill
+; LA32-NEXT:    fst.d $fs0, $sp, 16 # 8-byte Folded Spill
+; LA32-NEXT:    fst.d $fs1, $sp, 8 # 8-byte Folded Spill
+; LA32-NEXT:    fmov.s $fs0, $fa0
+; LA32-NEXT:    fmov.s $fa0, $fa1
+; LA32-NEXT:    bl %plt(__gnu_f2h_ieee)
+; LA32-NEXT:    bl %plt(__gnu_h2f_ieee)
+; LA32-NEXT:    fmov.s $fs1, $fa0
+; LA32-NEXT:    fmov.s $fa0, $fs0
+; LA32-NEXT:    bl %plt(__gnu_f2h_ieee)
+; LA32-NEXT:    bl %plt(__gnu_h2f_ieee)
+; LA32-NEXT:    fadd.s $fa0, $fa0, $fs1
+; LA32-NEXT:    fld.d $fs1, $sp, 8 # 8-byte Folded Reload
+; LA32-NEXT:    fld.d $fs0, $sp, 16 # 8-byte Folded Reload
+; LA32-NEXT:    ld.w $ra, $sp, 28 # 4-byte Folded Reload
+; LA32-NEXT:    addi.w $sp, $sp, 32
+; LA32-NEXT:    ret
+;
+; LA64-LABEL: test_fadd_reg:
+; LA64:       # %bb.0:
+; LA64-NEXT:    addi.d $sp, $sp, -32
+; LA64-NEXT:    st.d $ra, $sp, 24 # 8-byte Folded Spill
+; LA64-NEXT:    fst.d $fs0, $sp, 16 # 8-byte Folded Spill
+; LA64-NEXT:    fst.d $fs1, $sp, 8 # 8-byte Folded Spill
+; LA64-NEXT:    fmov.s $fs0, $fa0
+; LA64-NEXT:    fmov.s $fa0, $fa1
+; LA64-NEXT:    bl %plt(__gnu_f2h_ieee)
+; LA64-NEXT:    bl %plt(__gnu_h2f_ieee)
+; LA64-NEXT:    fmov.s $fs1, $fa0
+; LA64-NEXT:    fmov.s $fa0, $fs0
+; LA64-NEXT:    bl %plt(__gnu_f2h_ieee)
+; LA64-NEXT:    bl %plt(__gnu_h2f_ieee)
+; LA64-NEXT:    fadd.s $fa0, $fa0, $fs1
+; LA64-NEXT:    fld.d $fs1, $sp, 8 # 8-byte Folded Reload
+; LA64-NEXT:    fld.d $fs0, $sp, 16 # 8-byte Folded Reload
+; LA64-NEXT:    ld.d $ra, $sp, 24 # 8-byte Folded Reload
+; LA64-NEXT:    addi.d $sp, $sp, 32
+; LA64-NEXT:    ret
+  %r = fadd half %a, %b
+  ret half %r
+}
+
+define void @test_fadd_mem(ptr %p, ptr %q) nounwind {
+; LA32-LABEL: test_fadd_mem:
+; LA32:       # %bb.0:
+; LA32-NEXT:    addi.w $sp, $sp, -32
+; LA32-NEXT:    st.w $ra, $sp, 28 # 4-byte Folded Spill
+; LA32-NEXT:    st.w $fp, $sp, 24 # 4-byte Folded Spill
+; LA32-NEXT:    st.w $s0, $sp, 20 # 4-byte Folded Spill
+; LA32-NEXT:    fst.d $fs0, $sp, 8 # 8-byte Folded Spill
+; LA32-NEXT:    move $fp, $a1
+; LA32-NEXT:    move $s0, $a0
+; LA32-NEXT:    ld.hu $a0, $a0, 0
+; LA32-NEXT:    bl %plt(__gnu_h2f_ieee)
+; LA32-NEXT:    fmov.s $fs0, $fa0
+; LA32-NEXT:    ld.hu $a0, $fp, 0
+; LA32-NEXT:    bl %plt(__gnu_h2f_ieee)
+; LA32-NEXT:    fadd.s $fa0, $fs0, $fa0
+; LA32-NEXT:    bl %plt(__gnu_f2h_ieee)
+; LA32-NEXT:    st.h $a0, $s0, 0
+; LA32-NEXT:    fld.d $fs0, $sp, 8 # 8-byte Folded Reload
+; LA32-NEXT:    ld.w $s0, $sp, 20 # 4-byte Folded Reload
+; LA32-NEXT:    ld.w $fp, $sp, 24 # 4-byte Folded Reload
+; LA32-NEXT:    ld.w $ra, $sp, 28 # 4-byte Folded Reload
+; LA32-NEXT:    addi.w $sp, $sp, 32
+; LA32-NEXT:    ret
+;
+; LA64-LABEL: test_fadd_mem:
+; LA64:       # %bb.0:
+; LA64-NEXT:    addi.d $sp, $sp, -32
+; LA64-NEXT:    st.d $ra, $sp, 24 # 8-byte Folded Spill
+; LA64-NEXT:    st.d $fp, $sp, 16 # 8-byte Folded Spill
+; LA64-NEXT:    st.d $s0, $sp, 8 # 8-byte Folded Spill
+; LA64-NEXT:    fst.d $fs0, $sp, 0 # 8-byte Folded Spill
+; LA64-NEXT:    move $fp, $a1
+; LA64-NEXT:    move $s0, $a0
+; LA64-NEXT:    ld.hu $a0, $a0, 0
+; LA64-NEXT:    bl %plt(__gnu_h2f_ieee)
+; LA64-NEXT:    fmov.s $fs0, $fa0
+; LA64-NEXT:    ld.hu $a0, $fp, 0
+; LA64-NEXT:    bl %plt(__gnu_h2f_ieee)
+; LA64-NEXT:    fadd.s $fa0, $fs0, $fa0
+; LA64-NEXT:    bl %plt(__gnu_f2h_ieee)
+; LA64-NEXT:    st.h $a0, $s0, 0
+; LA64-NEXT:    fld.d $fs0, $sp, 0 # 8-byte Folded Reload
+; LA64-NEXT:    ld.d $s0, $sp, 8 # 8-byte Folded Reload
+; LA64-NEXT:    ld.d $fp, $sp, 16 # 8-byte Folded Reload
+; LA64-NEXT:    ld.d $ra, $sp, 24 # 8-byte Folded Reload
+; LA64-NEXT:    addi.d $sp, $sp, 32
+; LA64-NEXT:    ret
+  %a = load half, ptr %p
+  %b = load half, ptr %q
+  %r = fadd half %a, %b
+  store half %r, ptr %p
+  ret void
+}
+
+define half @test_fmul_reg(half %a, half %b) nounwind {
+; LA32-LABEL: test_fmul_reg:
+; LA32:       # %bb.0:
+; LA32-NEXT:    addi.w $sp, $sp, -32
+; LA32-NEXT:    st.w $ra, $sp, 28 # 4-byte Folded Spill
+; LA32-NEXT:    fst.d $fs0, $sp, 16 # 8-byte Folded Spill
+; LA32-NEXT:    fst.d $fs1, $sp, 8 # 8-byte Folded Spill
+; LA32-NEXT:    fmov.s $fs0, $fa0
+; LA32-NEXT:    fmov.s $fa0, $fa1
+; LA32-NEXT:    bl %plt(__gnu_f2h_ieee)
+; LA32-NEXT:    bl %plt(__gnu_h2f_ieee)
+; LA32-NEXT:    fmov.s $fs1, $fa0
+; LA32-NEXT:    fmov.s $fa0, $fs0
+; LA32-NEXT:    bl %plt(__gnu_f2h_ieee)
+; LA32-NEXT:    bl %plt(__gnu_h2f_ieee)
+; LA32-NEXT:    fmul.s $fa0, $fa0, $fs1
+; LA32-NEXT:    fld.d $fs1, $sp, 8 # 8-byte Folded Reload
+; LA32-NEXT:    fld.d $fs0, $sp, 16 # 8-byte Folded Reload
+; LA32-NEXT:    ld.w $ra, $sp, 28 # 4-byte Folded Reload
+; LA32-NEXT:    addi.w $sp, $sp, 32
+; LA32-NEXT:    ret
+;
+; LA64-LABEL: test_fmul_reg:
+; LA64:       # %bb.0:
+; LA64-NEXT:    addi.d $sp, $sp, -32
+; LA64-NEXT:    st.d $ra, $sp, 24 # 8-byte Folded Spill
+; LA64-NEXT:    fst.d $fs0, $sp, 16 # 8-byte Folded Spill
+; LA64-NEXT:    fst.d $fs1, $sp, 8 # 8-byte Folded Spill
+; LA64-NEXT:    fmov.s $fs0, $fa0
+; LA64-NEXT:    fmov.s $fa0, $fa1
+; LA64-NEXT:    bl %plt(__gnu_f2h_ieee)
+; LA64-NEXT:    bl %plt(__gnu_h2f_ieee)
+; LA64-NEXT:    fmov.s $fs1, $fa0
+; LA64-NEXT:    fmov.s $fa0, $fs0
+; LA64-NEXT:    bl %plt(__gnu_f2h_ieee)
+; LA64-NEXT:    bl %plt(__gnu_h2f_ieee)
+; LA64-NEXT:    fmul.s $fa0, $fa0, $fs1
+; LA64-NEXT:    fld.d $fs1, $sp, 8 # 8-byte Folded Reload
+; LA64-NEXT:    fld.d $fs0, $sp, 16 # 8-byte Folded Reload
+; LA64-NEXT:    ld.d $ra, $sp, 24 # 8-byte Folded Reload
+; LA64-NEXT:    addi.d $sp, $sp, 32
+; LA64-NEXT:    ret
+  %r = fmul half %a, %b
+  ret half %r
+}
+
+define void @test_fmul_mem(ptr %p, ptr %q) nounwind {
+; LA32-LABEL: test_fmul_mem:
+; LA32:       # %bb.0:
+; LA32-NEXT:    addi.w $sp, $sp, -32
+; LA32-NEXT:    st.w $ra, $sp, 28 # 4-byte Folded Spill
+; LA32-NEXT:    st.w $fp, $sp, 24 # 4-byte Folded Spill
+; LA32-NEXT:    st.w $s0, $sp, 20 # 4-byte Folded Spill
+; LA32-NEXT:    fst.d $fs0, $sp, 8 # 8-byte Folded Spill
+; LA32-NEXT:    move $fp, $a1
+; LA32-NEXT:    move $s0, $a0
+; LA32-NEXT:    ld.hu $a0, $a0, 0
+; LA32-NEXT:    bl %plt(__gnu_h2f_ieee)
+; LA32-NEXT:    fmov.s $fs0, $fa0
+; LA32-NEXT:    ld.hu $a0, $fp, 0
+; LA32-NEXT:    bl %plt(__gnu_h2f_ieee)
+; LA32-NEXT:    fmul.s $fa0, $fs0, $fa0
+; LA32-NEXT:    bl %plt(__gnu_f2h_ieee)
+; LA32-NEXT:    st.h $a0, $s0, 0
+; LA32-NEXT:    fld.d $fs0, $sp, 8 # 8-byte Folded Reload
+; LA32-NEXT:    ld.w $s0, $sp, 20 # 4-byte Folded Reload
+; LA32-NEXT:    ld.w $fp, $sp, 24 # 4-byte Folded Reload
+; LA32-NEXT:    ld.w $ra, $sp, 28 # 4-byte Folded Reload
+; LA32-NEXT:    addi.w $sp, $sp, 32
+; LA32-NEXT:    ret
+;
+; LA64-LABEL: test_fmul_mem:
+; LA64:       # %bb.0:
+; LA64-NEXT:    addi.d $sp, $sp, -32
+; LA64-NEXT:    st.d $ra, $sp, 24 # 8-byte Folded Spill
+; LA64-NEXT:    st.d $fp, $sp, 16 # 8-byte Folded Spill
+; LA64-NEXT:    st.d $s0, $sp, 8 # 8-byte Folded Spill
+; LA64-NEXT:    fst.d $fs0, $sp, 0 # 8-byte Folded Spill
+; LA64-NEXT:    move $fp, $a1
+; LA64-NEXT:    move $s0, $a0
+; LA64-NEXT:    ld.hu $a0, $a0, 0
+; LA64-NEXT:    bl %plt(__gnu_h2f_ieee)
+; LA64-NEXT:    fmov.s $fs0, $fa0
+; LA64-NEXT:    ld.hu $a0, $fp, 0
+; LA64-NEXT:    bl %plt(__gnu_h2f_ieee)
+; LA64-NEXT:    fmul.s $fa0, $fs0, $fa0
+; LA64-NEXT:    bl %plt(__gnu_f2h_ieee)
+; LA64-NEXT:    st.h $a0, $s0, 0
+; LA64-NEXT:    fld.d $fs0, $sp, 0 # 8-byte Folded Reload
+; LA64-NEXT:    ld.d $s0, $sp, 8 # 8-byte Folded Reload
+; LA64-NEXT:    ld.d $fp, $sp, 16 # 8-byte Folded Reload
+; LA64-NEXT:    ld.d $ra, $sp, 24 # 8-byte Folded Reload
+; LA64-NEXT:    addi.d $sp, $sp, 32
+; LA64-NEXT:    ret
+  %a = load half, ptr %p
+  %b = load half, ptr %q
+  %r = fmul half %a, %b
+  store half %r, ptr %p
+  ret void
+}

xen0n · 2024-06-05T10:52:27Z

cc @heiher @SixWeining

wangleiat

LGTM, thanks.

heiher

LGTM. Thanks.

Loongarch previously had a selection failure for `f16` math [1]. This was fixed in [2], which Rust got with the update to LLVM19. Enable it in `std/build.rs` so we start running tests. [1]: llvm/llvm-project#93894 [2]: llvm/llvm-project#94456

Loongarch previously had a selection failure for `f16` math [1]. This was fixed in [2], which Rust got with the update to LLVM 19 [3]. Enable loongarch in `std/build.rs` so we start running tests. [1]: llvm/llvm-project#93894 [2]: llvm/llvm-project#94456 [3]: rust-lang#127513

Enable `f16` tests on loongarch Loongarch previously had a selection failure for `f16` math llvm/llvm-project#93894. This was fixed in llvm/llvm-project#94456, which Rust got with the update to LLVM 19 rust-lang#127513. Enable loongarch in `std/build.rs` so we start running tests. try-job: dist-loongarch64-linux try-job: dist-loongarch64-musl

…ongarch (#107791) For zig with LLVM 19.1.0rc4, we are seeing the following error when bootstrapping a `loongarch64-linux-musl` target. ziglang/zig-bootstrap#164 (comment) It seems that this issue is caused by `PromoteFloatResult` is not handling FREEZE OP on loongarch. Here is the reproduction of the error: https://godbolt.org/z/PPfvWjjG5 ~~This patch adds the FREEZE OP handling with `PromoteFloatRes_UnaryOp` and adds a test case.~~ This patch changes loongarch's way of floating point promotion to soft promotion to avoid this problem. See: loongarch's handling of `half`: - #93894 - #94456 Also see: other float promotion FREEZE handling - 0019c2f

…ongarch (llvm#107791) For zig with LLVM 19.1.0rc4, we are seeing the following error when bootstrapping a `loongarch64-linux-musl` target. ziglang/zig-bootstrap#164 (comment) It seems that this issue is caused by `PromoteFloatResult` is not handling FREEZE OP on loongarch. Here is the reproduction of the error: https://godbolt.org/z/PPfvWjjG5 ~~This patch adds the FREEZE OP handling with `PromoteFloatRes_UnaryOp` and adds a test case.~~ This patch changes loongarch's way of floating point promotion to soft promotion to avoid this problem. See: loongarch's handling of `half`: - llvm#93894 - llvm#94456 Also see: other float promotion FREEZE handling - llvm@0019c2f (cherry picked from commit 13280d9)

[LoongArch] Allow f16 codegen with expansion to libcalls

bbb1854

The test case is adapted from llvm/test/CodeGen/RISCV/fp16-promote.ll, because it covers some more IR patterns that ought to be common. Fixes llvm#93894

llvmbot added the backend:loongarch label Jun 5, 2024

SixWeining requested review from SixWeining, heiher and wangleiat June 6, 2024 00:42

wangleiat approved these changes Jun 6, 2024

View reviewed changes

heiher approved these changes Jun 6, 2024

View reviewed changes

heiher merged commit d075b7b into llvm:main Jun 6, 2024
9 checks passed

xen0n deleted the issue93894 branch June 6, 2024 13:34

beetrees mentioned this pull request Jun 17, 2024

Update compiler_builtins to 0.1.114 rust-lang/rust#125016

Merged

beetrees mentioned this pull request Jul 7, 2024

LLVM miscompiles consecutive half operations by using too much precision on several backends #97975

Open

11 tasks

yxd-ym mentioned this pull request Aug 16, 2024

Content suggestion for This Week in LoongArch newsletter / 《每周一龙》新闻线索信箱 loongson-community/areweloongyet#16

Open

tgross35 mentioned this pull request Aug 22, 2024

Enable f16 tests on loongarch rust-lang/rust#129384

Closed

yxd-ym mentioned this pull request Sep 9, 2024

[loongarch][DAG][FREEZE] Fix crash when FREEZE a half(f16) type on loongarch #107791

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[LoongArch] Allow f16 codegen with expansion to libcalls #94456

[LoongArch] Allow f16 codegen with expansion to libcalls #94456

xen0n commented Jun 5, 2024

llvmbot commented Jun 5, 2024

xen0n commented Jun 5, 2024

wangleiat left a comment

heiher left a comment

[LoongArch] Allow f16 codegen with expansion to libcalls #94456

[LoongArch] Allow f16 codegen with expansion to libcalls #94456

Conversation

xen0n commented Jun 5, 2024

llvmbot commented Jun 5, 2024

xen0n commented Jun 5, 2024

wangleiat left a comment

Choose a reason for hiding this comment

heiher left a comment

Choose a reason for hiding this comment