Implement draw_arrays_instanced() in softpipe.

Modify the translate module to respect instance divisors and accept
instance id as a parameter to calculate input vertex offset.

3 files changed, 16 insertions, 3 deletions

diff --git a/src/gallium/auxiliary/translate/translate.h b/src/gallium/auxiliary/translate/translate.h
index 1afdf194b31..fb298471b8b 100644
--- a/src/gallium/auxiliary/translate/translate.h
+++ b/src/gallium/auxiliary/translate/translate.h
@@ -50,6 +50,7 @@ struct translate_element
    enum pipe_format output_format;
    unsigned input_buffer:8;
    unsigned input_offset:24;
+   unsigned instance_divisor;
    unsigned output_offset;
 };
 
@@ -79,6 +80,7 @@ struct translate {
    void (PIPE_CDECL *run)( struct translate *,
                            unsigned start,
                            unsigned count,
+                           unsigned instance_id,
                            void *output_buffer);
 };
 
diff --git a/src/gallium/auxiliary/translate/translate_generic.c b/src/gallium/auxiliary/translate/translate_generic.c
index 266e7ee81e6..0fa99274099 100644
--- a/src/gallium/auxiliary/translate/translate_generic.c
+++ b/src/gallium/auxiliary/translate/translate_generic.c
@@ -49,6 +49,7 @@ struct translate_generic {
       fetch_func fetch;
       unsigned buffer;
       unsigned input_offset;
+      unsigned instance_divisor;
 
       emit_func emit;
       unsigned output_offset;
@@ -607,6 +608,7 @@ static void PIPE_CDECL generic_run_elts( struct translate *translate,
 static void PIPE_CDECL generic_run( struct translate *translate,
                                     unsigned start,
                                     unsigned count,
+                                    unsigned instance_id,
                                     void *output_buffer )
 {
    struct translate_generic *tg = translate_generic(translate);
@@ -622,13 +624,20 @@ static void PIPE_CDECL generic_run( struct translate *translate,
 
       for (attr = 0; attr < nr_attrs; attr++) {
 	 float data[4];
-
-	 const char *src = (tg->attrib[attr].input_ptr + 
-			    tg->attrib[attr].input_stride * elt);
+         const char *src;
 
 	 char *dst = (vert + 
 		      tg->attrib[attr].output_offset);
 
+         if (tg->attrib[attr].instance_divisor) {
+            src = tg->attrib[attr].input_ptr +
+                  tg->attrib[attr].input_stride *
+                  (instance_id / tg->attrib[attr].instance_divisor);
+         } else {
+            src = tg->attrib[attr].input_ptr +
+                  tg->attrib[attr].input_stride * elt;
+         }
+
 	 tg->attrib[attr].fetch( src, data );
 
          if (0) debug_printf("vert %d attr %d: %f %f %f %f\n",
@@ -687,6 +696,7 @@ struct translate *translate_generic_create( const struct translate_key *key )
       tg->attrib[i].fetch = get_fetch_func(key->element[i].input_format);
       tg->attrib[i].buffer = key->element[i].input_buffer;
       tg->attrib[i].input_offset = key->element[i].input_offset;
+      tg->attrib[i].instance_divisor = key->element[i].instance_divisor;
 
       tg->attrib[i].emit = get_emit_func(key->element[i].output_format);
       tg->attrib[i].output_offset = key->element[i].output_offset;
diff --git a/src/gallium/auxiliary/translate/translate_sse.c b/src/gallium/auxiliary/translate/translate_sse.c
index b62db8d8f33..edd0be17f0b 100644
--- a/src/gallium/auxiliary/translate/translate_sse.c
+++ b/src/gallium/auxiliary/translate/translate_sse.c
@@ -637,6 +637,7 @@ static void PIPE_CDECL translate_sse_run_elts( struct translate *translate,
 static void PIPE_CDECL translate_sse_run( struct translate *translate,
 			 unsigned start,
 			 unsigned count,
+                         unsigned instance_id,
 			 void *output_buffer )
 {
    struct translate_sse *p = (struct translate_sse *)translate;