[PATCH] D157517: [AArch64][SVE] Add asm predicate constraint Uph

[Matt Devereau via Phabricator via llvm-commits]

MattDevereau created this revision.
MattDevereau added reviewers: sdesmalen, kmclaughlin.
Herald added subscribers: jdoerfert, psnobl, hiraditya, kristof.beyls.
Herald added a reviewer: efriedma.
Herald added a project: All.
MattDevereau requested review of this revision.
Herald added a project: LLVM.
Herald added a subscriber: llvm-commits.

Some instructions such as multi-vector LD1 only accept a range
of PN8-PN15 predicate-as-counter. This new constraint allows more
refined parsing and better decision making when parsing these
instructions from ASM, instead of defaulting to Upa which incorrectly
uses the whole range of registers P0-P15 from the register class PPR.


Repository:
  rG LLVM Github Monorepo

https://reviews.llvm.org/D157517

Files:
  llvm/docs/LangRef.rst
  llvm/lib/Target/AArch64/AArch64ISelLowering.cpp
  llvm/test/CodeGen/AArch64/aarch64-sve-asm.ll


Index: llvm/test/CodeGen/AArch64/aarch64-sve-asm.ll
===================================================================
--- llvm/test/CodeGen/AArch64/aarch64-sve-asm.ll
+++ llvm/test/CodeGen/AArch64/aarch64-sve-asm.ll
@@ -68,3 +68,14 @@
   %1 = tail call <vscale x 4 x i32> asm "incp $0.s, $1", "=w, at 3Upa,0"(<vscale x 16 x i1> %Pg, <vscale x 4 x i32> %Zn)
   ret <vscale x 4 x i32> %1
 }
+
+; Function Attrs: nounwind readnone
+; CHECK: [[ARG1:%[0-9]+]]:zpr = COPY $z1
+; CHECK: [[ARG2:%[0-9]+]]:zpr = COPY $z0
+; CHECK: [[ARG3:%[0-9]+]]:ppr = COPY $p0
+; CHECK: [[ARG4:%[0-9]+]]:ppr_p8to15 = COPY [[ARG3]]
+; CHECK: INLINEASM {{.*}} [[ARG4]]
+define <vscale x 8 x half> @foofoo(<vscale x 16 x i1> %Pg, <vscale x 8 x half> %Zn, <vscale x 8 x half> %Zm) {
+  %1 = tail call <vscale x 8 x half> asm "fadd $0.h, $1/m, $2.h, $3.h", "=w, at 3Uph,w,w"(<vscale x 16 x i1> %Pg, <vscale x 8 x half> %Zn, <vscale x 8 x half> %Zm)
+  ret <vscale x 8 x half> %1
+}
\ No newline at end of file
Index: llvm/lib/Target/AArch64/AArch64ISelLowering.cpp
===================================================================
--- llvm/lib/Target/AArch64/AArch64ISelLowering.cpp
+++ llvm/lib/Target/AArch64/AArch64ISelLowering.cpp
@@ -9978,19 +9978,33 @@
   return "r";
 }
 
-enum PredicateConstraint {
-  Upl,
-  Upa,
-  Invalid
-};
+enum PredicateConstraint { Uph, Upl, Upa, Invalid };
 
 static PredicateConstraint parsePredicateConstraint(StringRef Constraint) {
-  PredicateConstraint P = PredicateConstraint::Invalid;
-  if (Constraint == "Upa")
-    P = PredicateConstraint::Upa;
-  if (Constraint == "Upl")
-    P = PredicateConstraint::Upl;
-  return P;
+  if (Constraint == "Uph")
+    return PredicateConstraint::Uph;
+  else if (Constraint == "Upl")
+    return PredicateConstraint::Upl;
+  else if (Constraint == "Upa")
+    return PredicateConstraint::Upa;
+  return PredicateConstraint::Invalid;
+}
+
+static const TargetRegisterClass *
+getPredicateRegisterClass(PredicateConstraint Constraint, EVT VT) {
+  if (!VT.isScalableVector() || VT.getVectorElementType() != MVT::i1)
+    return nullptr;
+
+  switch (Constraint) {
+  default:
+    return nullptr;
+  case PredicateConstraint::Uph:
+    return &AArch64::PPR_p8to15RegClass;
+  case PredicateConstraint::Upl:
+    return &AArch64::PPR_3bRegClass;
+  case PredicateConstraint::Upa:
+    return &AArch64::PPRRegClass;
+  }
 }
 
 // The set of cc code supported is from
@@ -10182,13 +10196,8 @@
     }
   } else {
     PredicateConstraint PC = parsePredicateConstraint(Constraint);
-    if (PC != PredicateConstraint::Invalid) {
-      if (!VT.isScalableVector() || VT.getVectorElementType() != MVT::i1)
-        return std::make_pair(0U, nullptr);
-      bool restricted = (PC == PredicateConstraint::Upl);
-      return restricted ? std::make_pair(0U, &AArch64::PPR_3bRegClass)
-                        : std::make_pair(0U, &AArch64::PPRRegClass);
-    }
+    if (const TargetRegisterClass *RegClass = getPredicateRegisterClass(PC, VT))
+      return std::make_pair(0U, RegClass);
   }
   if (StringRef("{cc}").equals_insensitive(Constraint) ||
       parseConstraintCode(Constraint) != AArch64CC::Invalid)
Index: llvm/docs/LangRef.rst
===================================================================
--- llvm/docs/LangRef.rst
+++ llvm/docs/LangRef.rst
@@ -4994,6 +4994,7 @@
 - ``w``: A 32, 64, or 128-bit floating-point, SIMD or SVE vector register.
 - ``x``: Like w, but restricted to registers 0 to 15 inclusive.
 - ``y``: Like w, but restricted to SVE vector registers Z0 to Z7 inclusive.
+- ``Uph``: One of the high eight SVE predicate registers (P8 to P15)
 - ``Upl``: One of the low eight SVE predicate registers (P0 to P7)
 - ``Upa``: Any of the SVE predicate registers (P0 to P15)
 


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D157517.548641.patch
Type: text/x-patch
Size: 3742 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20230809/e3ce65f0/attachment.bin>