Remove the functinal form of 'assertOnGpu'

compiler/codegen/cg-expr.cpp

@@ -6288,9 +6288,6 @@ DEFINE_BASIC_PRIM(START_RMEM_FENCE)
 DEFINE_BASIC_PRIM(FINISH_RMEM_FENCE)
 
 DEFINE_PRIM(ASSERT_ON_GPU) {
-  // Remove the string argument to the GPU primitive.
-  call->get(1)->remove();
-
   ret = call->codegenBasicPrimitiveExpr();
 }
 

compiler/optimizations/gpuTransforms.cpp

@@ -414,13 +414,7 @@ class GpuAssertionReporter {
     if (assertion->isPrimitive(PRIM_ASSERT_GPU_ELIGIBLE)) {
       reason = "is marked with @gpu.assertEligible";
     } else {
-      INT_ASSERT(assertion->isPrimitive(PRIM_ASSERT_ON_GPU));
-      reason = "contains assertOnGpu()";
-      auto isAttributeSym = toSymExpr(assertion->get(1));
-      INT_ASSERT(isAttributeSym);
-      if (isAttributeSym->symbol() == gTrue) {
-        reason = "is marked with @assertOnGpu";
-      }
+      reason = "is marked with @assertOnGpu";
     }
     USR_FATAL_CONT(loc, "Loop %s but is not eligible for execution on a GPU", reason);
   }

modules/standard/AutoGpu.chpl

@@ -44,7 +44,7 @@ module AutoGpu {
                       "that the code can be executed on the GPU without ",
                       "runtime checks.");
     }
-    __primitive("chpl_assert_on_gpu", true);
+    __primitive("chpl_assert_on_gpu");
   }
 
   inline proc chpl__gpuBlockSizeAttr(param counter: int, arg: integral) {

modules/standard/GPU.chpl

@@ -133,18 +133,6 @@ module GPU
     gpuWrite((...args), "\n":c_ptrConst(c_char));
   }
 
-  /*
-    Will halt execution at runtime if called from outside a GPU.  If used on
-    first line in ``foreach`` or ``forall`` loop will also do a compile time
-    check that the loop is eligible for execution on a GPU.
-  */
-  pragma "insert line file info"
-  pragma "always propagate line file info"
-  @deprecated(notes="the functional form of assertOnGpu() is deprecated. Please use the @assertOnGpu loop attribute instead.")
-  inline proc assertOnGpu() {
-    __primitive("chpl_assert_on_gpu", false);
-  }
-
   /*
     Returns value of a per-multiprocessor counter that increments every clock cycle.
     This function is meant to be called to time sections of code within a GPU
@@ -227,7 +215,7 @@ module GPU
     pragma "codegen for GPU"
     extern "chpl_gpu_force_warp_sync" proc chpl_syncWarp(mask);
 
-    __primitive("chpl_assert_on_gpu", false);
+    __primitive("chpl_assert_on_gpu");
     chpl_syncWarp(mask);
   }
 
@@ -377,7 +365,7 @@ module GPU
     pragma "codegen for GPU"
     extern rtName proc chpl_atomicBinOp(x, val) : T;
 
-    __primitive("chpl_assert_on_gpu", false);
+    __primitive("chpl_assert_on_gpu");
     return chpl_atomicBinOp(c_ptrTo(x), val);
   }
 
@@ -388,7 +376,7 @@ module GPU
     pragma "codegen for GPU"
     extern rtName proc chpl_atomicTernOp(x, cmp, val) : T;
 
-    __primitive("chpl_assert_on_gpu", false);
+    __primitive("chpl_assert_on_gpu");
     return chpl_atomicTernOp(c_ptrTo(x), cmp, val);
   }
 

test/users/engin/context/gpuSharedMem.chpl

@@ -12,8 +12,8 @@ config const n = 20;
 on here.gpus[0] {
   var A : [0..<N] uint;
 
+  @assertOnGpu
   foreach i in simpleOneDim(N) {  // context should be coming from a new syntax
-    assertOnGpu();
     const context = new Context();
     const vectorContext = __primitive("outer context", context);