[flang][cuda] Update attribute compatibily check for unified matching…

… rule (#90679) This patch updates the compatibility checks for CUDA attribute iin preparation to implement the matching rules described in section 3.2.3. We this patch the compiler will still emit an error when there is multiple specific procedures that matches since the matching distances is not yet implemented. This will be done in a separate patch. https://docs.nvidia.com/hpc-sdk/archive/24.3/compilers/cuda-fortran-prog-guide/index.html#cfref-var-attr-unified-data gpu=unified and gpu=managed are not part of this patch since these options are not recognized by flang yet.
llvm · May 1, 2024 · 86e5d6f · 86e5d6f
1 parent ef1dbcd
commit 86e5d6f
Show file tree

Hide file tree

Showing 5 changed files with 43 additions and 9 deletions.
diff --git a/flang/include/flang/Common/Fortran.h b/flang/include/flang/Common/Fortran.h
@@ -114,8 +114,8 @@ static constexpr IgnoreTKRSet ignoreTKRAll{IgnoreTKR::Type, IgnoreTKR::Kind,
     IgnoreTKR::Rank, IgnoreTKR::Device, IgnoreTKR::Managed};
 std::string AsFortran(IgnoreTKRSet);
 
-bool AreCompatibleCUDADataAttrs(
-    std::optional<CUDADataAttr>, std::optional<CUDADataAttr>, IgnoreTKRSet);
+bool AreCompatibleCUDADataAttrs(std::optional<CUDADataAttr>,
+    std::optional<CUDADataAttr>, IgnoreTKRSet, bool allowUnifiedMatchingRule);
 
 static constexpr char blankCommonObjectName[] = "__BLNK__";
 

diff --git a/flang/lib/Common/Fortran.cpp b/flang/lib/Common/Fortran.cpp
@@ -97,8 +97,12 @@ std::string AsFortran(IgnoreTKRSet tkr) {
   return result;
 }
 
+/// Check compatibilty of CUDA attribute.
+/// When `allowUnifiedMatchingRule` is enabled, argument `x` represents the
+/// dummy argument attribute while `y` represents the actual argument attribute.
 bool AreCompatibleCUDADataAttrs(std::optional<CUDADataAttr> x,
-    std::optional<CUDADataAttr> y, IgnoreTKRSet ignoreTKR) {
+    std::optional<CUDADataAttr> y, IgnoreTKRSet ignoreTKR,
+    bool allowUnifiedMatchingRule) {
   if (!x && !y) {
     return true;
   } else if (x && y && *x == *y) {
@@ -114,6 +118,24 @@ bool AreCompatibleCUDADataAttrs(std::optional<CUDADataAttr> x,
       x.value_or(CUDADataAttr::Managed) == CUDADataAttr::Managed &&
       y.value_or(CUDADataAttr::Managed) == CUDADataAttr::Managed) {
     return true;
+  } else if (allowUnifiedMatchingRule) {
+    if (!x) { // Dummy argument has no attribute -> host
+      if (y && *y == CUDADataAttr::Managed || *y == CUDADataAttr::Unified) {
+        return true;
+      }
+    } else {
+      if (*x == CUDADataAttr::Device && y &&
+          (*y == CUDADataAttr::Managed || *y == CUDADataAttr::Unified)) {
+        return true;
+      } else if (*x == CUDADataAttr::Managed && y &&
+          *y == CUDADataAttr::Unified) {
+        return true;
+      } else if (*x == CUDADataAttr::Unified && y &&
+          *y == CUDADataAttr::Managed) {
+        return true;
+      }
+    }
+    return false;
   } else {
     return false;
   }

diff --git a/flang/lib/Evaluate/characteristics.cpp b/flang/lib/Evaluate/characteristics.cpp
@@ -362,8 +362,9 @@ bool DummyDataObject::IsCompatibleWith(const DummyDataObject &actual,
     }
   }
   if (!attrs.test(Attr::Value) &&
-      !common::AreCompatibleCUDADataAttrs(
-          cudaDataAttr, actual.cudaDataAttr, ignoreTKR)) {
+      !common::AreCompatibleCUDADataAttrs(cudaDataAttr, actual.cudaDataAttr,
+          ignoreTKR,
+          /*allowUnifiedMatchingRule=*/false)) {
     if (whyNot) {
       *whyNot = "incompatible CUDA data attributes";
     }
@@ -1754,8 +1755,9 @@ bool DistinguishUtils::Distinguishable(
   } else if (y.attrs.test(Attr::Allocatable) && x.attrs.test(Attr::Pointer) &&
       x.intent != common::Intent::In) {
     return true;
-  } else if (!common::AreCompatibleCUDADataAttrs(
-                 x.cudaDataAttr, y.cudaDataAttr, x.ignoreTKR | y.ignoreTKR)) {
+  } else if (!common::AreCompatibleCUDADataAttrs(x.cudaDataAttr, y.cudaDataAttr,
+                 x.ignoreTKR | y.ignoreTKR,
+                 /*allowUnifiedMatchingRule=*/false)) {
     return true;
   } else if (features_.IsEnabled(
                  common::LanguageFeature::DistinguishableSpecifics) &&

diff --git a/flang/lib/Semantics/check-call.cpp b/flang/lib/Semantics/check-call.cpp
@@ -897,8 +897,9 @@ static void CheckExplicitDataArg(const characteristics::DummyDataObject &dummy,
         actualDataAttr = common::CUDADataAttr::Device;
       }
     }
-    if (!common::AreCompatibleCUDADataAttrs(
-            dummyDataAttr, actualDataAttr, dummy.ignoreTKR)) {
+    if (!common::AreCompatibleCUDADataAttrs(dummyDataAttr, actualDataAttr,
+            dummy.ignoreTKR,
+            /*allowUnifiedMatchingRule=*/true)) {
       auto toStr{[](std::optional<common::CUDADataAttr> x) {
         return x ? "ATTRIBUTES("s +
                 parser::ToUpperCaseLetters(common::EnumToString(*x)) + ")"s

diff --git a/flang/test/Semantics/cuf13.cuf b/flang/test/Semantics/cuf13.cuf
@@ -6,6 +6,10 @@ module matching
     module procedure sub_device
   end interface
 
+  interface subman
+    module procedure sub_host
+  end interface
+
 contains
   subroutine sub_host(a)
     integer :: a(:)
@@ -21,8 +25,13 @@ program m
   use matching
 
   integer, pinned, allocatable :: a(:)
+  integer, managed, allocatable :: b(:)
   logical :: plog
   allocate(a(100), pinned = plog)
+  allocate(b(200))
 
   call sub(a)
+
+  call subman(b)
+
 end