r600: relax stride/alignment requirements for vertices

seems hw can do unaligned accesses and unaligned strides
removes extra conversion when using vbo's

however I needed to switch 3 component byte format to 4 component formats
for tests to pass. Somewhat sililar to GL_SHORT fix done earlier

removes assert and gains +2 piglit especially draw-vertices

src/mesa/drivers/dri/r600/r700_assembler.c

@@ -275,7 +275,10 @@ GLuint GetSurfaceFormat(GLenum eType, GLuint nChannels, GLuint * pClient_size)
                 case 2:
                     format = FMT_8_8; break;
                 case 3:
-                    format = FMT_8_8_8; break;
+                    /* for some (small/unaligned) strides using 4 comps works
+                     * better, probably same as GL_SHORT below
+                     * test piglit/draw-vertices */
+                    format = FMT_8_8_8_8; break;
                 case 4:
                     format = FMT_8_8_8_8; break;
                 default:

src/mesa/drivers/dri/r600/r700_render.c

@@ -625,11 +625,11 @@ static void r700SetupStreams(GLcontext *ctx, const struct gl_client_array *input
 
         stride = (input[i]->StrideB == 0) ? getTypeSize(input[i]->Type) * input[i]->Size : input[i]->StrideB;
 
-        if (input[i]->Type == GL_DOUBLE || input[i]->Type == GL_UNSIGNED_INT || input[i]->Type == GL_INT ||
+        if (input[i]->Type == GL_DOUBLE || input[i]->Type == GL_UNSIGNED_INT || input[i]->Type == GL_INT
 #if MESA_BIG_ENDIAN
-            getTypeSize(input[i]->Type) != 4 || 
+            || getTypeSize(input[i]->Type) != 4
 #endif
-            stride < 4) 
+            ) 
         {
             r700ConvertAttrib(ctx, count, input[i], &context->stream_desc[index]);
         } 
@@ -637,19 +637,10 @@ static void r700SetupStreams(GLcontext *ctx, const struct gl_client_array *input
         {
             if (input[i]->BufferObj->Name) 
             {
-                if (stride % 4 != 0) 
-                {
-                    assert(((intptr_t) input[i]->Ptr) % input[i]->StrideB == 0);
-                    r700AlignDataToDword(ctx, input[i], count, &context->stream_desc[index]);
-                    context->stream_desc[index].is_named_bo = GL_FALSE;
-                } 
-                else 
-                {
-                    context->stream_desc[index].stride = input[i]->StrideB;
-                    context->stream_desc[index].bo_offset = (intptr_t) input[i]->Ptr;
-                    context->stream_desc[index].bo = get_radeon_buffer_object(input[i]->BufferObj)->bo;
-                    context->stream_desc[index].is_named_bo = GL_TRUE;
-                }
+                context->stream_desc[index].stride = input[i]->StrideB;
+                context->stream_desc[index].bo_offset = (intptr_t) input[i]->Ptr;
+                context->stream_desc[index].bo = get_radeon_buffer_object(input[i]->BufferObj)->bo;
+                context->stream_desc[index].is_named_bo = GL_TRUE;
             } 
             else 
             {

src/mesa/drivers/dri/r600/r700_vertprog.c

@@ -461,11 +461,11 @@ static void r700TranslateAttrib(GLcontext *ctx, GLuint unLoc, int count, const s
 	stride = (input->StrideB == 0) ? getTypeSize(input->Type) * input->Size 
                                    : input->StrideB;
 
-    if (input->Type == GL_DOUBLE || input->Type == GL_UNSIGNED_INT || input->Type == GL_INT ||
+    if (input->Type == GL_DOUBLE || input->Type == GL_UNSIGNED_INT || input->Type == GL_INT
 #if MESA_BIG_ENDIAN
-        getTypeSize(input->Type) != 4 ||
+        || getTypeSize(input->Type) != 4
 #endif
-        stride < 4) 
+       ) 
     {
         pStreamDesc->type = GL_FLOAT;