Mesa (master): radeonsi: fix gl_BaseVertex in non-indexed draws

[Nicolai Hähnle]

Module: Mesa
Branch: master
Commit: 4f7e3fbb50f06ab523594a7ee4693b1a95c0e66a
URL:    http://cgit.freedesktop.org/mesa/mesa/commit/?id=4f7e3fbb50f06ab523594a7ee4693b1a95c0e66a

Author: Nicolai Hähnle <nicolai.haehnle at amd.com>
Date:   Wed Apr 12 11:01:19 2017 +0200

radeonsi: fix gl_BaseVertex in non-indexed draws

gl_BaseVertex is supposed to be 0 in non-indexed draws. Unfortunately, the
way they're implemented, the VGT always generates indices starting at 0,
and the VS prolog adds the start index.

There's a VGT_INDX_OFFSET register which causes the VGT to start at a
driver-defined index. However, this register cannot be written from
indirect draws.

So fix this unlikely case by setting a bit to tell the VS whether the
draw is indexed or not, so that gl_BaseVertex can be adjusted accordingly
when used.

Fixes a bug in
KHR-GL45.shader_draw_parameters_tests.ShaderMultiDrawArraysParameters.*

Reviewed-by: Marek Olšák <marek.olsak at amd.com>

---

 src/gallium/drivers/radeonsi/si_shader.c     | 17 +++++++++++++++--
 src/gallium/drivers/radeonsi/si_shader.h     |  2 ++
 src/gallium/drivers/radeonsi/si_state_draw.c |  8 ++++++--
 3 files changed, 23 insertions(+), 4 deletions(-)

diff --git a/src/gallium/drivers/radeonsi/si_shader.c b/src/gallium/drivers/radeonsi/si_shader.c
index 7adf76a0b0..0bda187cfd 100644
--- a/src/gallium/drivers/radeonsi/si_shader.c
+++ b/src/gallium/drivers/radeonsi/si_shader.c
@@ -1401,9 +1401,22 @@ static void declare_system_value(struct si_shader_context *ctx,
 		break;
 
 	case TGSI_SEMANTIC_BASEVERTEX:
-		value = LLVMGetParam(ctx->main_fn,
-				     SI_PARAM_BASE_VERTEX);
+	{
+		/* For non-indexed draws, the base vertex set by the driver
+		 * (for direct draws) or the CP (for indirect draws) is the
+		 * first vertex ID, but GLSL expects 0 to be returned.
+		 */
+		LLVMValueRef vs_state = LLVMGetParam(ctx->main_fn, SI_PARAM_VS_STATE_BITS);
+		LLVMValueRef indexed;
+
+		indexed = LLVMBuildLShr(gallivm->builder, vs_state, ctx->i32_1, "");
+		indexed = LLVMBuildTrunc(gallivm->builder, indexed, ctx->i1, "");
+
+		value = LLVMBuildSelect(gallivm->builder, indexed,
+					LLVMGetParam(ctx->main_fn, SI_PARAM_BASE_VERTEX),
+					ctx->i32_0, "");
 		break;
+	}
 
 	case TGSI_SEMANTIC_BASEINSTANCE:
 		value = LLVMGetParam(ctx->main_fn,
diff --git a/src/gallium/drivers/radeonsi/si_shader.h b/src/gallium/drivers/radeonsi/si_shader.h
index fdb0dd482b..f145eab14f 100644
--- a/src/gallium/drivers/radeonsi/si_shader.h
+++ b/src/gallium/drivers/radeonsi/si_shader.h
@@ -220,6 +220,8 @@ enum {
 /* Clamp vertex color output (only used in VS as VS). */
 #define S_VS_STATE_CLAMP_VERTEX_COLOR(x)	(((unsigned)(x) & 0x1) << 0)
 #define C_VS_STATE_CLAMP_VERTEX_COLOR		0xFFFFFFFE
+#define S_VS_STATE_INDEXED(x)			(((unsigned)(x) & 0x1) << 1)
+#define C_VS_STATE_INDEXED			0xFFFFFFFD
 #define S_VS_STATE_LS_OUT_PATCH_SIZE(x)		(((unsigned)(x) & 0x1FFF) << 8)
 #define C_VS_STATE_LS_OUT_PATCH_SIZE		0xFFE000FF
 #define S_VS_STATE_LS_OUT_VERTEX_SIZE(x)	(((unsigned)(x) & 0xFF) << 24)
diff --git a/src/gallium/drivers/radeonsi/si_state_draw.c b/src/gallium/drivers/radeonsi/si_state_draw.c
index 0d70ea9d6d..aa528ce439 100644
--- a/src/gallium/drivers/radeonsi/si_state_draw.c
+++ b/src/gallium/drivers/radeonsi/si_state_draw.c
@@ -494,8 +494,12 @@ static void si_emit_rasterizer_prim_state(struct si_context *sctx)
 	sctx->last_sc_line_stipple = rs->pa_sc_line_stipple;
 }
 
-static void si_emit_vs_state(struct si_context *sctx)
+static void si_emit_vs_state(struct si_context *sctx,
+			     const struct pipe_draw_info *info)
 {
+	sctx->current_vs_state &= C_VS_STATE_INDEXED;
+	sctx->current_vs_state |= S_VS_STATE_INDEXED(!!info->indexed);
+
 	if (sctx->current_vs_state != sctx->last_vs_state) {
 		struct radeon_winsys_cs *cs = sctx->b.gfx.cs;
 
@@ -1305,7 +1309,7 @@ void si_draw_vbo(struct pipe_context *ctx, const struct pipe_draw_info *info)
 	si_emit_rasterizer_prim_state(sctx);
 	if (sctx->tes_shader.cso)
 		si_emit_derived_tess_state(sctx, info, &num_patches);
-	si_emit_vs_state(sctx);
+	si_emit_vs_state(sctx, info);
 	si_emit_draw_registers(sctx, info, num_patches);
 
 	si_ce_pre_draw_synchronization(sctx);