From: Artem Belevich <tra@google.com>
Date: Fri, 23 Mar 2018 19:49:03 +0000 (+0000)
Subject: [CUDA] Fixed false error reporting in case of calling H->G->HD->D.
X-Git-Tag: llvmorg-7.0.0-rc1~9788
X-Git-Url: http://review.tizen.org/git/?a=commitdiff_plain;h=e2ae8b55108c08aa0718a020355de83e10897103;p=platform%2Fupstream%2Fllvm.git

[CUDA] Fixed false error reporting in case of calling H->G->HD->D.

Launching a kernel from the host code does not generate code for the
kernel itself. This fixes an issue with clang erroneously reporting
an error for a HD->D call from within the kernel.

Differential Revision: https://reviews.llvm.org/D44837

llvm-svn: 328362
---

diff --git a/clang/lib/Sema/SemaCUDA.cpp b/clang/lib/Sema/SemaCUDA.cpp
index cac5f68..ccd93fa 100644
--- a/clang/lib/Sema/SemaCUDA.cpp
+++ b/clang/lib/Sema/SemaCUDA.cpp
@@ -790,9 +790,12 @@ bool Sema::CheckCUDACall(SourceLocation Loc, FunctionDecl *Callee) {
   // If the caller is known-emitted, mark the callee as known-emitted.
   // Otherwise, mark the call in our call graph so we can traverse it later.
   bool CallerKnownEmitted = IsKnownEmitted(*this, Caller);
-  if (CallerKnownEmitted)
-    MarkKnownEmitted(*this, Caller, Callee, Loc);
-  else {
+  if (CallerKnownEmitted) {
+    // Host-side references to a __global__ function refer to the stub, so the
+    // function itself is never emitted and therefore should not be marked.
+    if (getLangOpts().CUDAIsDevice || IdentifyCUDATarget(Callee) != CFT_Global)
+      MarkKnownEmitted(*this, Caller, Callee, Loc);
+  } else {
     // If we have
     //   host fn calls kernel fn calls host+device,
     // the HD function does not get instantiated on the host.  We model this by
diff --git a/clang/test/SemaCUDA/call-device-fn-from-host.cu b/clang/test/SemaCUDA/call-device-fn-from-host.cu
index 3ea013f..26215d5 100644
--- a/clang/test/SemaCUDA/call-device-fn-from-host.cu
+++ b/clang/test/SemaCUDA/call-device-fn-from-host.cu
@@ -83,3 +83,10 @@ template <typename T>
 __host__ __device__ void fn_ptr_template() {
   auto* ptr = &device_fn;  // Not an error because the template isn't instantiated.
 }
+
+// Launching a kernel from a host function does not result in code generation
+// for it, so calling HD function which calls a D function should not trigger
+// errors.
+static __host__ __device__ void hd_func() { device_fn(); }
+__global__ void kernel() { hd_func(); }
+void host_func(void) { kernel<<<1, 1>>>(); }