Minor documentation fixes

Author: njroussel

include/drjit-core/jit.h

@@ -1073,12 +1073,12 @@ extern JIT_EXPORT uint32_t jit_var_mem_map(JIT_ENUM JitBackend backend,
  * Copy a memory region onto the device and return its variable index. Its
  * reference count is initialized to \c 1.
  *
+ * \param backend
+ *    The JIT backend in which the variable should be created
+ *
  * \param atype
  *    Enumeration characterizing the "flavor" of the source memory.
  *
- * \param cuda
- *    Is this a CUDA variable?
- *
  * \param vtype
  *    Type of the variable to be created, see \ref VarType for details.
  *

src/util.h

@@ -64,7 +64,7 @@ extern uint32_t jitc_mkperm(JitBackend backend, const uint32_t *values, uint32_t
 /// Perform a synchronous copy operation
 extern void jitc_memcpy(JitBackend backend, void *dst, const void *src, size_t size);
 
-/// Perform an assynchronous copy operation
+/// Perform an asynchronous copy operation
 extern void jitc_memcpy_async(JitBackend backend, void *dst, const void *src, size_t size);
 
 /// Asynchronously update a single element in memory

src/var.cpp

@@ -1531,11 +1531,9 @@ uint32_t jitc_var_mem_copy(JitBackend backend, AllocType atype, VarType vtype,
     size_t total_size = (size_t) size * (size_t) type_size[(int) vtype];
     void *target_ptr;
 
-    ThreadState *ts = thread_state(backend);
     if (backend == JitBackend::CUDA) {
         target_ptr = jitc_malloc(AllocType::Device, total_size);
 
-        scoped_set_context guard(ts->context);
         if (atype == AllocType::HostAsync) {
             jitc_fail("jit_var_mem_copy(): copy from HostAsync to GPU memory not supported!");
         } else if (atype == AllocType::Host) {