Merge branch 'master' into ci-mi100

CHANGELOG.md

@@ -26,6 +26,8 @@ and this project adheres to [Semantic Versioning](http://semver.org/spec/v2.0.0.
 
 * Interpreter no longer crashes on attributes in patterns.
 
+* Fixes to array indexing through C API when using GPU backends.
+
 ## [0.25.19]
 
 ### Added

docs/c-api.rst

@@ -392,7 +392,7 @@ tuple is simply a record with numeric fields.
 
 The projection and construction functions are equivalent in
 functionality to writing entry points by hand, and so serve only to
-cut down on boilerplate.  Important things to be aware of:
+cut down on boilerplate. Important things to be aware of:
 
 1. The objects constructed though these functions have their own
    lifetime (like any objects returned from an entry point) and must
@@ -405,6 +405,8 @@ cut down on boilerplate.  Important things to be aware of:
    always, you don't have to worry about this if you never write entry
    points that consume their arguments.
 
+3. You must synchronise before using any scalar results.
+
 The precise functions generated depend on the fields of the record.
 The following functions assume a record with Futhark-level type ``type
 t = {foo: t1, bar: t2}`` where ``t1`` and ``t2`` are also opaque

src/Futhark/CodeGen/Backends/GenericC/Types.hs

@@ -552,7 +552,7 @@ recordArrayIndexFunctions space _types desc rank elemtype vds = do
           -- variable.
           copy
             CopyNoBarrier
-            [C.cexp|&v->$id:(tupleField j)|]
+            [C.cexp|(unsigned char*)&v->$id:(tupleField j)|]
             [C.cexp|0|]
             DefaultSpace
             [C.cexp|arr->$id:(tupleField j)->mem.mem|]
@@ -571,7 +571,7 @@ recordArrayIndexFunctions space _types desc rank elemtype vds = do
             CopyNoBarrier
             [C.cexp|v->$id:(tupleField j)->mem.mem|]
             [C.cexp|0|]
-            DefaultSpace
+            space
             [C.cexp|arr->$id:(tupleField j)->mem.mem|]
             (indexExp pt r [C.cexp|arr->$id:(tupleField j)->shape|])
             space

tests_lib/c/test_opaque_array.c

@@ -50,6 +50,7 @@ void test1(struct futhark_context *ctx) {
 
     float out1_host[2];
     assert(futhark_values_f32_1d(ctx, out1, out1_host) == 0);
+    assert(futhark_context_sync(ctx) == 0);
     assert(memcmp(out1_host, b, sizeof(float)*2) == 0);
 
     assert(futhark_free_opaque_7e1a81531cc3c23bb9093ed822aec320(ctx, out) == 0);

tests_lib/c/test_record_array.c

@@ -40,21 +40,25 @@ void test1(struct futhark_context *ctx) {
   // Test indexing: in bounds.
   struct futhark_opaque_tup2_i32_tup2_f32_f32* trip_fut;
   assert(futhark_index_opaque_arr1d_tup2_i32_tup2_f32_f32(ctx, &trip_fut, a_b_b_fut, 1) == 0);
+  assert(futhark_context_sync(ctx) == 0); // XXX, would be nice if this was not required.
   struct futhark_opaque_tup2_f32_f32* pair_fut;
   assert(futhark_project_opaque_tup2_i32_tup2_f32_f32_1(ctx, &pair_fut, trip_fut) == 0);
   {
     int x;
     assert(futhark_project_opaque_tup2_i32_tup2_f32_f32_0(ctx, &x, trip_fut) == 0);
+    assert(futhark_context_sync(ctx) == 0);
     assert(x == a[1]);
   }
   {
     float x;
     assert(futhark_project_opaque_tup2_f32_f32_0(ctx, &x, pair_fut) == 0);
+    assert(futhark_context_sync(ctx) == 0);
     assert(x == b[1]);
   }
   {
     float x;
     assert(futhark_project_opaque_tup2_f32_f32_1(ctx, &x, pair_fut) == 0);
+    assert(futhark_context_sync(ctx) == 0);
     assert(x == b[1]);
   }