intel: Rewrite the world of push/pull params

This moves us away to the array of pointers model and onto a model where
each param is represented by a generic uint32_t handle.  We reserve 2^16
of these handles for builtins that get generated by somewhere inside the
compiler and have well-defined meanings.  Generic params have handles
whose meanings are defined by the driver.

The primary downside to this new approach is that it moves a little bit
of the work that we would normally do at compile time to draw time.  On
my laptop this hurts OglBatch6 by no more than 1% and doesn't seem to
have any measurable affect on OglBatch7.  So, while this may come back
to bite us, it doesn't look too bad.

Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

src/intel/vulkan/anv_cmd_buffer.c

@@ -629,6 +629,26 @@ anv_cmd_buffer_merge_dynamic(struct anv_cmd_buffer *cmd_buffer,
    return state;
 }
 
+static uint32_t
+anv_push_constant_value(struct anv_push_constants *data, uint32_t param)
+{
+   if (BRW_PARAM_IS_BUILTIN(param)) {
+      switch (param) {
+      case BRW_PARAM_BUILTIN_ZERO:
+         return 0;
+      default:
+         unreachable("Invalid param builtin");
+      }
+   } else {
+      uint32_t offset = ANV_PARAM_PUSH_OFFSET(param);
+      assert(offset % sizeof(uint32_t) == 0);
+      if (offset < data->size)
+         return *(uint32_t *)((uint8_t *)data + offset);
+      else
+         return 0;
+   }
+}
+
 struct anv_state
 anv_cmd_buffer_push_constants(struct anv_cmd_buffer *cmd_buffer,
                               gl_shader_stage stage)
@@ -653,10 +673,8 @@ anv_cmd_buffer_push_constants(struct anv_cmd_buffer *cmd_buffer,
 
    /* Walk through the param array and fill the buffer with data */
    uint32_t *u32_map = state.map;
-   for (unsigned i = 0; i < prog_data->nr_params; i++) {
-      uint32_t offset = (uintptr_t)prog_data->param[i];
-      u32_map[i] = *(uint32_t *)((uint8_t *)data + offset);
-   }
+   for (unsigned i = 0; i < prog_data->nr_params; i++)
+      u32_map[i] = anv_push_constant_value(data, prog_data->param[i]);
 
    anv_state_flush(cmd_buffer->device, state);
 
@@ -695,8 +713,7 @@ anv_cmd_buffer_cs_push_constants(struct anv_cmd_buffer *cmd_buffer)
       for (unsigned i = 0;
            i < cs_prog_data->push.cross_thread.dwords;
            i++) {
-         uint32_t offset = (uintptr_t)prog_data->param[i];
-         u32_map[i] = *(uint32_t *)((uint8_t *)data + offset);
+         u32_map[i] = anv_push_constant_value(data, prog_data->param[i]);
       }
    }
 
@@ -708,8 +725,8 @@ anv_cmd_buffer_cs_push_constants(struct anv_cmd_buffer *cmd_buffer)
          unsigned src = cs_prog_data->push.cross_thread.dwords;
          for ( ; src < prog_data->nr_params; src++, dst++) {
             if (src != cs_prog_data->thread_local_id_index) {
-               uint32_t offset = (uintptr_t)prog_data->param[src];
-               u32_map[dst] = *(uint32_t *)((uint8_t *)data + offset);
+               u32_map[dst] =
+                  anv_push_constant_value(data, prog_data->param[src]);
             } else {
                u32_map[dst] = t * cs_prog_data->simd_size;
             }