From c51f133197437d01696abd9513fbcda4b16b897c Mon Sep 17 00:00:00 2001
From: =?utf8?q?Kristian=20H=C3=B8gsberg=20Kristensen?= <krh@bitplanet.net>
Date: Fri, 11 Dec 2015 11:18:18 -0800
Subject: [PATCH] i965: Move brw_cs_fill_local_id_payload() to libi965_compiler

This is a helper function for setting up the local invocation ID
payload according to the cs_prog_data generated by the compiler. It's
intended to be available to users of libi965_compiler so move it there.
---
 src/mesa/drivers/dri/i965/brw_compiler.h |  7 +++++++
 src/mesa/drivers/dri/i965/brw_cs.c       | 36 --------------------------------
 src/mesa/drivers/dri/i965/brw_cs.h       |  4 ----
 src/mesa/drivers/dri/i965/brw_fs.cpp     | 36 ++++++++++++++++++++++++++++++++
 4 files changed, 43 insertions(+), 40 deletions(-)

diff --git a/src/mesa/drivers/dri/i965/brw_compiler.h b/src/mesa/drivers/dri/i965/brw_compiler.h
index 218d9c7..514788c 100644
--- a/src/mesa/drivers/dri/i965/brw_compiler.h
+++ b/src/mesa/drivers/dri/i965/brw_compiler.h
@@ -701,6 +701,13 @@ brw_compile_cs(const struct brw_compiler *compiler, void *log_data,
                unsigned *final_assembly_size,
                char **error_str);
 
+/**
+ * Fill out local id payload for compute shader according to cs_prog_data.
+ */
+void
+brw_cs_fill_local_id_payload(const struct brw_cs_prog_data *cs_prog_data,
+                             void *buffer, uint32_t threads, uint32_t stride);
+
 #ifdef __cplusplus
 } /* extern "C" */
 #endif
diff --git a/src/mesa/drivers/dri/i965/brw_cs.c b/src/mesa/drivers/dri/i965/brw_cs.c
index d88e822..9eadb7e 100644
--- a/src/mesa/drivers/dri/i965/brw_cs.c
+++ b/src/mesa/drivers/dri/i965/brw_cs.c
@@ -34,42 +34,6 @@
 #include "brw_program.h"
 #include "glsl/ir_uniform.h"
 
-void
-brw_cs_fill_local_id_payload(const struct brw_cs_prog_data *prog_data,
-                             void *buffer, uint32_t threads, uint32_t stride)
-{
-   if (prog_data->local_invocation_id_regs == 0)
-      return;
-
-   /* 'stride' should be an integer number of registers, that is, a multiple
-    * of 32 bytes.
-    */
-   assert(stride % 32 == 0);
-
-   unsigned x = 0, y = 0, z = 0;
-   for (unsigned t = 0; t < threads; t++) {
-      uint32_t *param = (uint32_t *) buffer + stride * t / 4;
-
-      for (unsigned i = 0; i < prog_data->simd_size; i++) {
-         param[0 * prog_data->simd_size + i] = x;
-         param[1 * prog_data->simd_size + i] = y;
-         param[2 * prog_data->simd_size + i] = z;
-
-         x++;
-         if (x == prog_data->local_size[0]) {
-            x = 0;
-            y++;
-            if (y == prog_data->local_size[1]) {
-               y = 0;
-               z++;
-               if (z == prog_data->local_size[2])
-                  z = 0;
-            }
-         }
-      }
-   }
-}
-
 static void
 assign_cs_binding_table_offsets(const struct brw_device_info *devinfo,
                                 const struct gl_shader_program *shader_prog,
diff --git a/src/mesa/drivers/dri/i965/brw_cs.h b/src/mesa/drivers/dri/i965/brw_cs.h
index 9ce39fb..890a0c8 100644
--- a/src/mesa/drivers/dri/i965/brw_cs.h
+++ b/src/mesa/drivers/dri/i965/brw_cs.h
@@ -32,10 +32,6 @@ extern "C" {
 void
 brw_upload_cs_prog(struct brw_context *brw);
 
-void
-brw_cs_fill_local_id_payload(const struct brw_cs_prog_data *cs_prog_data,
-                             void *buffer, uint32_t threads, uint32_t stride);
-
 #ifdef __cplusplus
 }
 #endif
diff --git a/src/mesa/drivers/dri/i965/brw_fs.cpp b/src/mesa/drivers/dri/i965/brw_fs.cpp
index 5e8acec..c833ef0 100644
--- a/src/mesa/drivers/dri/i965/brw_fs.cpp
+++ b/src/mesa/drivers/dri/i965/brw_fs.cpp
@@ -5690,3 +5690,39 @@ brw_compile_cs(const struct brw_compiler *compiler, void *log_data,
 
    return g.get_assembly(final_assembly_size);
 }
+
+void
+brw_cs_fill_local_id_payload(const struct brw_cs_prog_data *prog_data,
+                             void *buffer, uint32_t threads, uint32_t stride)
+{
+   if (prog_data->local_invocation_id_regs == 0)
+      return;
+
+   /* 'stride' should be an integer number of registers, that is, a multiple
+    * of 32 bytes.
+    */
+   assert(stride % 32 == 0);
+
+   unsigned x = 0, y = 0, z = 0;
+   for (unsigned t = 0; t < threads; t++) {
+      uint32_t *param = (uint32_t *) buffer + stride * t / 4;
+
+      for (unsigned i = 0; i < prog_data->simd_size; i++) {
+         param[0 * prog_data->simd_size + i] = x;
+         param[1 * prog_data->simd_size + i] = y;
+         param[2 * prog_data->simd_size + i] = z;
+
+         x++;
+         if (x == prog_data->local_size[0]) {
+            x = 0;
+            y++;
+            if (y == prog_data->local_size[1]) {
+               y = 0;
+               z++;
+               if (z == prog_data->local_size[2])
+                  z = 0;
+            }
+         }
+      }
+   }
+}
-- 
2.7.4