[llvm] r354846 - [WebAssembly] Properly align fp128 arguments in outgoing varargs arguments

[Dan Gohman via llvm-commits]

Author: djg
Date: Mon Feb 25 21:20:19 2019
New Revision: 354846

URL: http://llvm.org/viewvc/llvm-project?rev=354846&view=rev
Log:
[WebAssembly] Properly align fp128 arguments in outgoing varargs arguments

For outgoing varargs arguments, it's necessary to check the OrigAlign field
of the corresponding OutputArg entry to determine argument alignment, rather
than just computing an alignment from the argument value type. This is
because types like fp128 are split into multiple argument values, with
narrower types that don't reflect the ABI alignment of the full fp128.

This fixes the printf("printfL: %4.*Lf\n", 2, lval); testcase.

Differential Revision: https://reviews.llvm.org/D58656

Modified:
    llvm/trunk/lib/Target/WebAssembly/WebAssemblyISelLowering.cpp
    llvm/trunk/test/CodeGen/WebAssembly/varargs.ll

Modified: llvm/trunk/lib/Target/WebAssembly/WebAssemblyISelLowering.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/WebAssembly/WebAssemblyISelLowering.cpp?rev=354846&r1=354845&r2=354846&view=diff
==============================================================================
--- llvm/trunk/lib/Target/WebAssembly/WebAssemblyISelLowering.cpp (original)
+++ llvm/trunk/lib/Target/WebAssembly/WebAssemblyISelLowering.cpp Mon Feb 25 21:20:19 2019
@@ -674,13 +674,16 @@ WebAssemblyTargetLowering::LowerCall(Cal
   if (IsVarArg) {
     // Outgoing non-fixed arguments are placed in a buffer. First
     // compute their offsets and the total amount of buffer space needed.
-    for (SDValue Arg :
-         make_range(OutVals.begin() + NumFixedArgs, OutVals.end())) {
+    for (unsigned I = NumFixedArgs; I < Outs.size(); ++I) {
+      const ISD::OutputArg &Out = Outs[I];
+      SDValue &Arg = OutVals[I];
       EVT VT = Arg.getValueType();
       assert(VT != MVT::iPTR && "Legalized args should be concrete");
       Type *Ty = VT.getTypeForEVT(*DAG.getContext());
+      unsigned Align = std::max(Out.Flags.getOrigAlign(),
+                                Layout.getABITypeAlignment(Ty));
       unsigned Offset = CCInfo.AllocateStack(Layout.getTypeAllocSize(Ty),
-                                             Layout.getABITypeAlignment(Ty));
+                                             Align);
       CCInfo.addLoc(CCValAssign::getMem(ArgLocs.size(), VT.getSimpleVT(),
                                         Offset, VT.getSimpleVT(),
                                         CCValAssign::Full));

Modified: llvm/trunk/test/CodeGen/WebAssembly/varargs.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/WebAssembly/varargs.ll?rev=354846&r1=354845&r2=354846&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/WebAssembly/varargs.ll (original)
+++ llvm/trunk/test/CodeGen/WebAssembly/varargs.ll Mon Feb 25 21:20:19 2019
@@ -163,6 +163,32 @@ define void @nonlegal_fixed(fp128 %x, ..
   ret void
 }
 
+; Test that an fp128 argument is properly aligned and allocated
+; within a vararg buffer.
+
+; CHECK-LABEL: call_fp128_alignment:
+; CHECK:      global.get      $push7=, __stack_pointer
+; CHECK-NEXT: i32.const       $push8=, 32
+; CHECK-NEXT: i32.sub         $push12=, $pop7, $pop8
+; CHECK-NEXT: local.tee       $push11=, $1=, $pop12
+; CHECK-NEXT: global.set      __stack_pointer, $pop11
+; CHECK-NEXT: i32.const       $push0=, 24
+; CHECK-NEXT: i32.add         $push1=, $1, $pop0
+; CHECK-NEXT: i64.const       $push2=, -9223372036854775808
+; CHECK-NEXT: i64.store       0($pop1), $pop2
+; CHECK-NEXT: i32.const       $push3=, 16
+; CHECK-NEXT: i32.add         $push4=, $1, $pop3
+; CHECK-NEXT: i64.const       $push5=, 1
+; CHECK-NEXT: i64.store       0($pop4), $pop5
+; CHECK-NEXT: i32.const       $push6=, 7
+; CHECK-NEXT: i32.store       0($1), $pop6
+; CHECK-NEXT: call            callee, $1
+define void @call_fp128_alignment(i8* %p) {
+entry:
+  call void (...) @callee(i8 7, fp128 0xL00000000000000018000000000000000)
+  ret void
+}
+
 declare void @llvm.va_start(i8*)
 declare void @llvm.va_end(i8*)
 declare void @llvm.va_copy(i8*, i8*)