Merge branch 'gallium-no-rhw-position'

Conflicts:
	src/gallium/drivers/r300/r300_blit.c
	src/gallium/drivers/r300/r300_emit.c
	src/gallium/drivers/r300/r300_state_derived.c

15 files changed, 210 insertions, 331 deletions

diff --git a/src/gallium/drivers/nv30/nv30_context.h b/src/gallium/drivers/nv30/nv30_context.h
index ca3d6aca7f..ea259aadf3 100644
--- a/src/gallium/drivers/nv30/nv30_context.h
+++ b/src/gallium/drivers/nv30/nv30_context.h
@@ -101,7 +101,6 @@ struct nv30_blend_state {
 struct nv30_state {
 	unsigned scissor_enabled;
 	unsigned stipple_enabled;
-	unsigned viewport_bypass;
 	unsigned fp_samplers;
 
 	uint64_t dirty;
diff --git a/src/gallium/drivers/nv30/nv30_state_viewport.c b/src/gallium/drivers/nv30/nv30_state_viewport.c
index 2d7781292b..6fccd6b60e 100644
--- a/src/gallium/drivers/nv30/nv30_state_viewport.c
+++ b/src/gallium/drivers/nv30/nv30_state_viewport.c
@@ -5,55 +5,25 @@ nv30_state_viewport_validate(struct nv30_context *nv30)
 {
 	struct pipe_viewport_state *vpt = &nv30->viewport;
 	struct nouveau_stateobj *so;
-	unsigned bypass;
-
-	if (/*nv30->render_mode == HW &&*/
-	    !nv30->rasterizer->pipe.bypass_vs_clip_and_viewport)
-		bypass = 0;
-	else
-		bypass = 1;
 
 	if (nv30->state.hw[NV30_STATE_VIEWPORT] &&
-	    (bypass || !(nv30->dirty & NV30_NEW_VIEWPORT)) &&
-	    nv30->state.viewport_bypass == bypass)
+	    !(nv30->dirty & NV30_NEW_VIEWPORT))
 		return FALSE;
-	nv30->state.viewport_bypass = bypass;
 
 	so = so_new(3, 10, 0);
-	if (!bypass) {
-		so_method(so, nv30->screen->rankine,
-			  NV34TCL_VIEWPORT_TRANSLATE_X, 8);
-		so_data  (so, fui(vpt->translate[0]));
-		so_data  (so, fui(vpt->translate[1]));
-		so_data  (so, fui(vpt->translate[2]));
-		so_data  (so, fui(vpt->translate[3]));
-		so_data  (so, fui(vpt->scale[0]));
-		so_data  (so, fui(vpt->scale[1]));
-		so_data  (so, fui(vpt->scale[2]));
-		so_data  (so, fui(vpt->scale[3]));
-/*		so_method(so, nv30->screen->rankine, 0x1d78, 1);
-		so_data  (so, 1);
-*/	} else {
-		so_method(so, nv30->screen->rankine,
-			  NV34TCL_VIEWPORT_TRANSLATE_X, 8);
-		so_data  (so, fui(0.0));
-		so_data  (so, fui(0.0));
-		so_data  (so, fui(0.0));
-		so_data  (so, fui(0.0));
-		so_data  (so, fui(1.0));
-		so_data  (so, fui(1.0));
-		so_data  (so, fui(1.0));
-		so_data  (so, fui(0.0));
-		/* Not entirely certain what this is yet.  The DDX uses this
-		 * value also as it fixes rendering when you pass
-		 * pre-transformed vertices to the GPU.  My best gusss is that
-		 * this bypasses some culling/clipping stage.  Might be worth
-		 * noting that points/lines are uneffected by whatever this
-		 * value fixes, only filled polygons are effected.
-		 */
-/*		so_method(so, nv30->screen->rankine, 0x1d78, 1);
-		so_data  (so, 0x110);
-*/	}
+	so_method(so, nv30->screen->rankine,
+		  NV34TCL_VIEWPORT_TRANSLATE_X, 8);
+	so_data  (so, fui(vpt->translate[0]));
+	so_data  (so, fui(vpt->translate[1]));
+	so_data  (so, fui(vpt->translate[2]));
+	so_data  (so, fui(vpt->translate[3]));
+	so_data  (so, fui(vpt->scale[0]));
+	so_data  (so, fui(vpt->scale[1]));
+	so_data  (so, fui(vpt->scale[2]));
+	so_data  (so, fui(vpt->scale[3]));
+/*	so_method(so, nv30->screen->rankine, 0x1d78, 1);
+	so_data  (so, 1);
+*/
 	/* TODO/FIXME: never saw value 0x0110 in renouveau dumps, only 0x0001 */
 	so_method(so, nv30->screen->rankine, 0x1d78, 1);
 	so_data  (so, 1);
diff --git a/src/gallium/drivers/nv40/nv40_context.h b/src/gallium/drivers/nv40/nv40_context.h
index 4861924dac..97fb6a2ef9 100644
--- a/src/gallium/drivers/nv40/nv40_context.h
+++ b/src/gallium/drivers/nv40/nv40_context.h
@@ -101,7 +101,6 @@ struct nv40_blend_state {
 struct nv40_state {
 	unsigned scissor_enabled;
 	unsigned stipple_enabled;
-	unsigned viewport_bypass;
 	unsigned fp_samplers;
 
 	uint64_t dirty;
diff --git a/src/gallium/drivers/nv40/nv40_state_viewport.c b/src/gallium/drivers/nv40/nv40_state_viewport.c
index 9919ba1d0b..3aacb00f99 100644
--- a/src/gallium/drivers/nv40/nv40_state_viewport.c
+++ b/src/gallium/drivers/nv40/nv40_state_viewport.c
@@ -5,55 +5,24 @@ nv40_state_viewport_validate(struct nv40_context *nv40)
 {
 	struct pipe_viewport_state *vpt = &nv40->viewport;
 	struct nouveau_stateobj *so;
-	unsigned bypass;
-
-	if (nv40->render_mode == HW &&
-	    !nv40->rasterizer->pipe.bypass_vs_clip_and_viewport)
-		bypass = 0;
-	else
-		bypass = 1;
 
 	if (nv40->state.hw[NV40_STATE_VIEWPORT] &&
-	    (bypass || !(nv40->dirty & NV40_NEW_VIEWPORT)) &&
-	    nv40->state.viewport_bypass == bypass)
+	    !(nv40->dirty & NV40_NEW_VIEWPORT))
 		return FALSE;
-	nv40->state.viewport_bypass = bypass;
 
 	so = so_new(2, 9, 0);
-	if (!bypass) {
-		so_method(so, nv40->screen->curie,
-			  NV40TCL_VIEWPORT_TRANSLATE_X, 8);
-		so_data  (so, fui(vpt->translate[0]));
-		so_data  (so, fui(vpt->translate[1]));
-		so_data  (so, fui(vpt->translate[2]));
-		so_data  (so, fui(vpt->translate[3]));
-		so_data  (so, fui(vpt->scale[0]));
-		so_data  (so, fui(vpt->scale[1]));
-		so_data  (so, fui(vpt->scale[2]));
-		so_data  (so, fui(vpt->scale[3]));
-		so_method(so, nv40->screen->curie, 0x1d78, 1);
-		so_data  (so, 1);
-	} else {
-		so_method(so, nv40->screen->curie,
-			  NV40TCL_VIEWPORT_TRANSLATE_X, 8);
-		so_data  (so, fui(0.0));
-		so_data  (so, fui(0.0));
-		so_data  (so, fui(0.0));
-		so_data  (so, fui(0.0));
-		so_data  (so, fui(1.0));
-		so_data  (so, fui(1.0));
-		so_data  (so, fui(1.0));
-		so_data  (so, fui(0.0));
-		/* Not entirely certain what this is yet.  The DDX uses this
-		 * value also as it fixes rendering when you pass
-		 * pre-transformed vertices to the GPU.  My best gusss is that
-		 * this bypasses some culling/clipping stage.  Might be worth
-		 * noting that points/lines are uneffected by whatever this
-		 * value fixes, only filled polygons are effected.
-		 */
-		so_method(so, nv40->screen->curie, 0x1d78, 1);
-		so_data  (so, 0x110);
-	}
+	so_method(so, nv40->screen->curie,
+		  NV40TCL_VIEWPORT_TRANSLATE_X, 8);
+	so_data  (so, fui(vpt->translate[0]));
+	so_data  (so, fui(vpt->translate[1]));
+	so_data  (so, fui(vpt->translate[2]));
+	so_data  (so, fui(vpt->translate[3]));
+	so_data  (so, fui(vpt->scale[0]));
+	so_data  (so, fui(vpt->scale[1]));
+	so_data  (so, fui(vpt->scale[2]));
+	so_data  (so, fui(vpt->scale[3]));
+	so_method(so, nv40->screen->curie, 0x1d78, 1);
+	so_data  (so, 1);
 
 	so_ref(so, &nv40->state.hw[NV40_STATE_VIEWPORT]);
 	so_ref(NULL, &so);
diff --git a/src/gallium/drivers/nv50/nv50_context.h b/src/gallium/drivers/nv50/nv50_context.h
index b4de3e2ba5..c540594b94 100644
--- a/src/gallium/drivers/nv50/nv50_context.h
+++ b/src/gallium/drivers/nv50/nv50_context.h
@@ -127,7 +127,6 @@ struct nv50_state {
 	struct nouveau_stateobj *scissor;
 	unsigned scissor_enabled;
 	struct nouveau_stateobj *viewport;
-	unsigned viewport_bypass;
 	struct nouveau_stateobj *tsc_upload;
 	struct nouveau_stateobj *tic_upload;
 	unsigned miptree_nr[PIPE_SHADER_TYPES];
diff --git a/src/gallium/drivers/nv50/nv50_state_validate.c b/src/gallium/drivers/nv50/nv50_state_validate.c
index efab94cab7..a91b31015e 100644
--- a/src/gallium/drivers/nv50/nv50_state_validate.c
+++ b/src/gallium/drivers/nv50/nv50_state_validate.c
@@ -375,50 +375,32 @@ nv50_state_validate(struct nv50_context *nv50)
 scissor_uptodate:
 
 	if (nv50->dirty & (NV50_NEW_VIEWPORT | NV50_NEW_RASTERIZER)) {
-		unsigned bypass;
-
-		if (!nv50->rasterizer->pipe.bypass_vs_clip_and_viewport)
-			bypass = 0;
-		else
-			bypass = 1;
-
 		if (nv50->state.viewport &&
-		    (bypass || !(nv50->dirty & NV50_NEW_VIEWPORT)) &&
-		    nv50->state.viewport_bypass == bypass)
+		    !(nv50->dirty & NV50_NEW_VIEWPORT))
 			goto viewport_uptodate;
-		nv50->state.viewport_bypass = bypass;
 
 		so = so_new(5, 9, 0);
-		if (!bypass) {
-			so_method(so, tesla, NV50TCL_VIEWPORT_TRANSLATE_X(0), 3);
-			so_data  (so, fui(nv50->viewport.translate[0]));
-			so_data  (so, fui(nv50->viewport.translate[1]));
-			so_data  (so, fui(nv50->viewport.translate[2]));
-			so_method(so, tesla, NV50TCL_VIEWPORT_SCALE_X(0), 3);
-			so_data  (so, fui(nv50->viewport.scale[0]));
-			so_data  (so, fui(nv50->viewport.scale[1]));
-			so_data  (so, fui(nv50->viewport.scale[2]));
-
-			so_method(so, tesla, NV50TCL_VIEWPORT_TRANSFORM_EN, 1);
-			so_data  (so, 1);
-			/* 0x0000 = remove whole primitive only (xyz)
-			 * 0x1018 = remove whole primitive only (xy), clamp z
-			 * 0x1080 = clip primitive (xyz)
-			 * 0x1098 = clip primitive (xy), clamp z
-			 */
-			so_method(so, tesla, NV50TCL_VIEW_VOLUME_CLIP_CTRL, 1);
-			so_data  (so, 0x1080);
-			/* no idea what 0f90 does */
-			so_method(so, tesla, 0x0f90, 1);
-			so_data  (so, 0);
-		} else {
-			so_method(so, tesla, NV50TCL_VIEWPORT_TRANSFORM_EN, 1);
-			so_data  (so, 0);
-			so_method(so, tesla, NV50TCL_VIEW_VOLUME_CLIP_CTRL, 1);
-			so_data  (so, 0x0000);
-			so_method(so, tesla, 0x0f90, 1);
-			so_data  (so, 1);
-		}
+		so_method(so, tesla, NV50TCL_VIEWPORT_TRANSLATE_X(0), 3);
+		so_data  (so, fui(nv50->viewport.translate[0]));
+		so_data  (so, fui(nv50->viewport.translate[1]));
+		so_data  (so, fui(nv50->viewport.translate[2]));
+		so_method(so, tesla, NV50TCL_VIEWPORT_SCALE_X(0), 3);
+		so_data  (so, fui(nv50->viewport.scale[0]));
+		so_data  (so, fui(nv50->viewport.scale[1]));
+		so_data  (so, fui(nv50->viewport.scale[2]));
+
+		so_method(so, tesla, NV50TCL_VIEWPORT_TRANSFORM_EN, 1);
+		so_data  (so, 1);
+		/* 0x0000 = remove whole primitive only (xyz)
+		 * 0x1018 = remove whole primitive only (xy), clamp z
+		 * 0x1080 = clip primitive (xyz)
+		 * 0x1098 = clip primitive (xy), clamp z
+		 */
+		so_method(so, tesla, NV50TCL_VIEW_VOLUME_CLIP_CTRL, 1);
+		so_data  (so, 0x1080);
+		/* no idea what 0f90 does */
+		so_method(so, tesla, 0x0f90, 1);
+		so_data  (so, 0);
 
 		so_ref(so, &nv50->state.viewport);
 		so_ref(NULL, &so);
diff --git a/src/gallium/drivers/r300/r300_blit.c b/src/gallium/drivers/r300/r300_blit.c
index 32d05749bd..8acb1098b9 100644
--- a/src/gallium/drivers/r300/r300_blit.c
+++ b/src/gallium/drivers/r300/r300_blit.c
@@ -34,6 +34,8 @@ static void r300_blitter_save_states(struct r300_context* r300)
     util_blitter_save_rasterizer(r300->blitter, r300->rs_state.state);
     util_blitter_save_fragment_shader(r300->blitter, r300->fs);
     util_blitter_save_vertex_shader(r300->blitter, r300->vs_state.state);
+    util_blitter_save_viewport(r300->blitter, &r300->viewport);
+    util_blitter_save_clip(r300->blitter, &r300->clip);
 }
 
 /* Clear currently bound buffers. */
diff --git a/src/gallium/drivers/r300/r300_context.h b/src/gallium/drivers/r300/r300_context.h
index b98fe347b8..0d1518a05b 100644
--- a/src/gallium/drivers/r300/r300_context.h
+++ b/src/gallium/drivers/r300/r300_context.h
@@ -356,12 +356,14 @@ struct r300_context {
 
     struct pipe_stencil_ref stencil_ref;
 
+    struct pipe_clip_state clip;
+
+    struct pipe_viewport_state viewport;
+
     /* Bitmask of dirty state objects. */
     uint32_t dirty_state;
     /* Flag indicating whether or not the HW is dirty. */
     uint32_t dirty_hw;
-    /* Whether the TCL engine should be in bypass mode. */
-    boolean tcl_bypass;
     /* Whether polygon offset is enabled. */
     boolean polygon_offset_enabled;
     /* Z buffer bit depth. */
diff --git a/src/gallium/drivers/r300/r300_emit.c b/src/gallium/drivers/r300/r300_emit.c
index 17d55ba3b4..addb28bded 100644
--- a/src/gallium/drivers/r300/r300_emit.c
+++ b/src/gallium/drivers/r300/r300_emit.c
@@ -175,23 +175,15 @@ static const float * get_shader_constant(
                     break;
 
                 case RC_STATE_R300_VIEWPORT_SCALE:
-                    if (r300->tcl_bypass) {
-                        vec[0] = 1;
-                        vec[1] = 1;
-                        vec[2] = 1;
-                    } else {
-                        vec[0] = viewport->xscale;
-                        vec[1] = viewport->yscale;
-                        vec[2] = viewport->zscale;
-                    }
+                    vec[0] = viewport->xscale;
+                    vec[1] = viewport->yscale;
+                    vec[2] = viewport->zscale;
                     break;
 
                 case RC_STATE_R300_VIEWPORT_OFFSET:
-                    if (!r300->tcl_bypass) {
-                        vec[0] = viewport->xoffset;
-                        vec[1] = viewport->yoffset;
-                        vec[2] = viewport->zoffset;
-                    }
+                    vec[0] = viewport->xoffset;
+                    vec[1] = viewport->yoffset;
+                    vec[2] = viewport->zoffset;
                     break;
 
                 default:
@@ -946,22 +938,16 @@ void r300_emit_viewport_state(struct r300_context* r300,
     struct r300_viewport_state* viewport = (struct r300_viewport_state*)state;
     CS_LOCALS(r300);
 
-    if (r300->tcl_bypass) {
-        BEGIN_CS(2); /* XXX tcl_bypass will be removed in gallium anyway */
-        OUT_CS_REG(R300_VAP_VTE_CNTL, 0);
-        END_CS;
-    } else {
-        BEGIN_CS(size);
-        OUT_CS_REG_SEQ(R300_SE_VPORT_XSCALE, 6);
-        OUT_CS_32F(viewport->xscale);
-        OUT_CS_32F(viewport->xoffset);
-        OUT_CS_32F(viewport->yscale);
-        OUT_CS_32F(viewport->yoffset);
-        OUT_CS_32F(viewport->zscale);
-        OUT_CS_32F(viewport->zoffset);
-        OUT_CS_REG(R300_VAP_VTE_CNTL, viewport->vte_control);
-        END_CS;
-    }
+     BEGIN_CS(size);
+     OUT_CS_REG_SEQ(R300_SE_VPORT_XSCALE, 6);
+     OUT_CS_32F(viewport->xscale);
+     OUT_CS_32F(viewport->xoffset);
+     OUT_CS_32F(viewport->yscale);
+     OUT_CS_32F(viewport->yoffset);
+     OUT_CS_32F(viewport->zscale);
+     OUT_CS_32F(viewport->zoffset);
+     OUT_CS_REG(R300_VAP_VTE_CNTL, viewport->vte_control);
+     END_CS;
 }
 
 void r300_emit_ztop_state(struct r300_context* r300,
diff --git a/src/gallium/drivers/r300/r300_state.c b/src/gallium/drivers/r300/r300_state.c
index 12bf083871..b732380a14 100644
--- a/src/gallium/drivers/r300/r300_state.c
+++ b/src/gallium/drivers/r300/r300_state.c
@@ -376,6 +376,8 @@ static void r300_set_clip_state(struct pipe_context* pipe,
 {
     struct r300_context* r300 = r300_context(pipe);
 
+    r300->clip = *state;
+
     if (r300_screen(pipe->screen)->caps->has_tcl) {
         memcpy(r300->clip_state.state, state, sizeof(struct pipe_clip_state));
         r300->clip_state.size = 29;
@@ -714,8 +716,7 @@ static void* r300_create_rs_state(struct pipe_context* pipe,
 
     /* If bypassing TCL, or if no TCL engine is present, turn off the HW TCL.
      * Else, enable HW TCL and force Draw's TCL off. */
-    if (state->bypass_vs_clip_and_viewport ||
-            !r300screen->caps->has_tcl) {
+    if (!r300screen->caps->has_tcl) {
         rs->vap_control_status |= R300_VAP_TCL_BYPASS;
     }
 
@@ -811,11 +812,9 @@ static void r300_bind_rs_state(struct pipe_context* pipe, void* state)
     }
 
     if (rs) {
-        r300->tcl_bypass = rs->rs.bypass_vs_clip_and_viewport;
         r300->polygon_offset_enabled = rs->rs.offset_cw || rs->rs.offset_ccw;
         r300->rs_state.dirty = TRUE;
     } else {
-        r300->tcl_bypass = FALSE;
         r300->polygon_offset_enabled = FALSE;
     }
 
@@ -982,6 +981,8 @@ static void r300_set_viewport_state(struct pipe_context* pipe,
     struct r300_viewport_state* viewport =
         (struct r300_viewport_state*)r300->viewport_state.state;
 
+    r300->viewport = *state;
+
     /* Do the transform in HW. */
     viewport->vte_control = R300_VTX_W0_FMT;
 
diff --git a/src/gallium/drivers/r300/r300_state_derived.c b/src/gallium/drivers/r300/r300_state_derived.c
index 6eb7f2bfd1..9c8e907fdf 100644
--- a/src/gallium/drivers/r300/r300_state_derived.c
+++ b/src/gallium/drivers/r300/r300_state_derived.c
@@ -119,13 +119,7 @@ static void r300_vertex_psc(struct r300_context* r300)
 
     memset(vformat, 0, sizeof(struct r300_vertex_stream_state));
 
-    /* If TCL is bypassed, map vertex streams to equivalent VS output
-     * locations. */
-    if (r300->tcl_bypass) {
-        stream_tab = vs->stream_loc_notcl;
-    } else {
-        stream_tab = identity;
-    }
+    stream_tab = identity;
 
     /* Vertex shaders have no semantics on their inputs,
      * so PSC should just route stuff based on the vertex elements,
diff --git a/src/gallium/drivers/softpipe/sp_video_context.c b/src/gallium/drivers/softpipe/sp_video_context.c
index 272ad40a90..242aaac466 100644
--- a/src/gallium/drivers/softpipe/sp_video_context.c
+++ b/src/gallium/drivers/softpipe/sp_video_context.c
@@ -176,7 +176,6 @@ init_pipe_state(struct sp_mpeg12_context *ctx)
    rast.line_stipple_factor = 0;
    rast.line_stipple_pattern = 0;
    rast.line_last_pixel = 0;
-   rast.bypass_vs_clip_and_viewport = 0;
    rast.line_width = 1;
    rast.point_smooth = 0;
    rast.point_quad_rasterization = 0;
diff --git a/src/gallium/drivers/svga/svga_state_framebuffer.c b/src/gallium/drivers/svga/svga_state_framebuffer.c
index b4cafb8f21..4d0c06bd77 100644
--- a/src/gallium/drivers/svga/svga_state_framebuffer.c
+++ b/src/gallium/drivers/svga/svga_state_framebuffer.c
@@ -120,174 +120,153 @@ static int emit_viewport( struct svga_context *svga,
    float fb_width = svga->curr.framebuffer.width;
    float fb_height = svga->curr.framebuffer.height;
 
-   memset( &prescale, 0, sizeof(prescale) );
-
-   if (svga->curr.rast->templ.bypass_vs_clip_and_viewport) {
-
-      /* Avoid POSITIONT as it has a non trivial implementation outside the D3D
-       * API. Always generate a vertex shader.
-       */
-      rect.x = 0;
-      rect.y = 0;
-      rect.w = svga->curr.framebuffer.width;
-      rect.h = svga->curr.framebuffer.height;
-
-      prescale.scale[0] = 2.0 / (float)rect.w;
-      prescale.scale[1] = - 2.0 / (float)rect.h;
-      prescale.scale[2] = 1.0;
-      prescale.scale[3] = 1.0;
-      prescale.translate[0] = -1.0f;
-      prescale.translate[1] = 1.0f;
-      prescale.translate[2] = 0;
-      prescale.translate[3] = 0;
-      prescale.enabled = TRUE;
-   } else {
-
-      /* Examine gallium viewport transformation and produce a screen
-       * rectangle and possibly vertex shader pre-transformation to
-       * get the same results.
-       */
-      float fx =        viewport->scale[0] * -1.0 + viewport->translate[0];
-      float fy = flip * viewport->scale[1] * -1.0 + viewport->translate[1];
-      float fw =        viewport->scale[0] * 2; 
-      float fh = flip * viewport->scale[1] * 2; 
-
-      SVGA_DBG(DEBUG_VIEWPORT,
-               "\ninitial %f,%f %fx%f\n",
-               fx,
-               fy,
-               fw,
-               fh);
-
-      prescale.scale[0] = 1.0;
-      prescale.scale[1] = 1.0;
-      prescale.scale[2] = 1.0;
-      prescale.scale[3] = 1.0;
-      prescale.translate[0] = 0;
-      prescale.translate[1] = 0;
-      prescale.translate[2] = 0;
-      prescale.translate[3] = 0;
-      prescale.enabled = TRUE;
-
-
-
-      if (fw < 0) {
-         prescale.scale[0] *= -1.0;
-         prescale.translate[0] += -fw;
-         fw = -fw;
-         fx =        viewport->scale[0] * 1.0 + viewport->translate[0];
-      }
+   float fx =        viewport->scale[0] * -1.0 + viewport->translate[0];
+   float fy = flip * viewport->scale[1] * -1.0 + viewport->translate[1];
+   float fw =        viewport->scale[0] * 2; 
+   float fh = flip * viewport->scale[1] * 2; 
 
-      if (fh < 0) {
-         prescale.scale[1] *= -1.0;
-         prescale.translate[1] += -fh;
-         fh = -fh;
-         fy = flip * viewport->scale[1] * 1.0 + viewport->translate[1];
-      }
+   memset( &prescale, 0, sizeof(prescale) );
 
-      if (fx < 0) {
-         prescale.translate[0] += fx;
-         prescale.scale[0] *= fw / (fw + fx); 
-         fw += fx;
-         fx = 0;
-      }
+   /* Examine gallium viewport transformation and produce a screen
+    * rectangle and possibly vertex shader pre-transformation to
+    * get the same results.
+    */
 
-      if (fy < 0) {
-         prescale.translate[1] += fy;
-         prescale.scale[1] *= fh / (fh + fy); 
-         fh += fy;
-         fy = 0;
-      }
+   SVGA_DBG(DEBUG_VIEWPORT,
+            "\ninitial %f,%f %fx%f\n",
+            fx,
+            fy,
+            fw,
+            fh);
+
+   prescale.scale[0] = 1.0;
+   prescale.scale[1] = 1.0;
+   prescale.scale[2] = 1.0;
+   prescale.scale[3] = 1.0;
+   prescale.translate[0] = 0;
+   prescale.translate[1] = 0;
+   prescale.translate[2] = 0;
+   prescale.translate[3] = 0;
+   prescale.enabled = TRUE;
+
+
+
+   if (fw < 0) {
+      prescale.scale[0] *= -1.0;
+      prescale.translate[0] += -fw;
+      fw = -fw;
+      fx =        viewport->scale[0] * 1.0 + viewport->translate[0];
+   }
 
-      if (fx + fw > fb_width) {
-         prescale.scale[0] *= fw / (fb_width - fx); 
-         prescale.translate[0] -= fx * (fw / (fb_width - fx));
-         prescale.translate[0] += fx;
-         fw = fb_width - fx;
-         
-      }
+   if (fh < 0) {
+      prescale.scale[1] *= -1.0;
+      prescale.translate[1] += -fh;
+      fh = -fh;
+      fy = flip * viewport->scale[1] * 1.0 + viewport->translate[1];
+   }
 
-      if (fy + fh > fb_height) {
-         prescale.scale[1] *= fh / (fb_height - fy);
-         prescale.translate[1] -= fy * (fh / (fb_height - fy));
-         prescale.translate[1] += fy;
-         fh = fb_height - fy;
-      }
+   if (fx < 0) {
+      prescale.translate[0] += fx;
+      prescale.scale[0] *= fw / (fw + fx); 
+      fw += fx;
+      fx = 0;
+   }
 
-      if (fw < 0 || fh < 0) {
-         fw = fh = fx = fy = 0;
-         degenerate = TRUE;
-         goto out;
-      }
+   if (fy < 0) {
+      prescale.translate[1] += fy;
+      prescale.scale[1] *= fh / (fh + fy); 
+      fh += fy;
+      fy = 0;
+   }
 
+   if (fx + fw > fb_width) {
+      prescale.scale[0] *= fw / (fb_width - fx); 
+      prescale.translate[0] -= fx * (fw / (fb_width - fx));
+      prescale.translate[0] += fx;
+      fw = fb_width - fx;
+      
+   }
 
-      /* D3D viewport is integer space.  Convert fx,fy,etc. to
-       * integers.
-       *
-       * TODO: adjust pretranslate correct for any subpixel error
-       * introduced converting to integers.
-       */
-      rect.x = fx;
-      rect.y = fy;
-      rect.w = fw;
-      rect.h = fh;
+   if (fy + fh > fb_height) {
+      prescale.scale[1] *= fh / (fb_height - fy);
+      prescale.translate[1] -= fy * (fh / (fb_height - fy));
+      prescale.translate[1] += fy;
+      fh = fb_height - fy;
+   }
 
-      SVGA_DBG(DEBUG_VIEWPORT,
-               "viewport error %f,%f %fx%f\n",
-               fabs((float)rect.x - fx),
-               fabs((float)rect.y - fy),
-               fabs((float)rect.w - fw),
-               fabs((float)rect.h - fh));
+   if (fw < 0 || fh < 0) {
+      fw = fh = fx = fy = 0;
+      degenerate = TRUE;
+      goto out;
+   }
 
-      SVGA_DBG(DEBUG_VIEWPORT,
-               "viewport %d,%d %dx%d\n",
-               rect.x,
-               rect.y,
-               rect.w,
-               rect.h);
-      
 
-      /* Finally, to get GL rasterization rules, need to tweak the
-       * screen-space coordinates slightly relative to D3D which is
-       * what hardware implements natively.
-       */
-      if (svga->curr.rast->templ.gl_rasterization_rules) {
-         float adjust_x = 0.0;
-         float adjust_y = 0.0;
-
-         switch (svga->curr.reduced_prim) {
-         case PIPE_PRIM_LINES:
-            adjust_x = -0.5;
-            adjust_y = 0;
-            break;
-         case PIPE_PRIM_POINTS:
-         case PIPE_PRIM_TRIANGLES:
-            adjust_x = -0.375;
-            adjust_y = -0.5;
-            break;
-         }
-
-         prescale.translate[0] += adjust_x;
-         prescale.translate[1] += adjust_y;
-         prescale.translate[2] = 0.5; /* D3D clip space */
-         prescale.scale[2]     = 0.5; /* D3D clip space */
+   /* D3D viewport is integer space.  Convert fx,fy,etc. to
+    * integers.
+    *
+    * TODO: adjust pretranslate correct for any subpixel error
+    * introduced converting to integers.
+    */
+   rect.x = fx;
+   rect.y = fy;
+   rect.w = fw;
+   rect.h = fh;
+
+   SVGA_DBG(DEBUG_VIEWPORT,
+            "viewport error %f,%f %fx%f\n",
+            fabs((float)rect.x - fx),
+            fabs((float)rect.y - fy),
+            fabs((float)rect.w - fw),
+            fabs((float)rect.h - fh));
+
+   SVGA_DBG(DEBUG_VIEWPORT,
+            "viewport %d,%d %dx%d\n",
+            rect.x,
+            rect.y,
+            rect.w,
+            rect.h);
+
+
+   /* Finally, to get GL rasterization rules, need to tweak the
+    * screen-space coordinates slightly relative to D3D which is
+    * what hardware implements natively.
+    */
+   if (svga->curr.rast->templ.gl_rasterization_rules) {
+      float adjust_x = 0.0;
+      float adjust_y = 0.0;
+
+      switch (svga->curr.reduced_prim) {
+      case PIPE_PRIM_LINES:
+         adjust_x = -0.5;
+         adjust_y = 0;
+         break;
+      case PIPE_PRIM_POINTS:
+      case PIPE_PRIM_TRIANGLES:
+         adjust_x = -0.375;
+         adjust_y = -0.5;
+         break;
       }
 
+      prescale.translate[0] += adjust_x;
+      prescale.translate[1] += adjust_y;
+      prescale.translate[2] = 0.5; /* D3D clip space */
+      prescale.scale[2]     = 0.5; /* D3D clip space */
+   }
 
-      range_min = viewport->scale[2] * -1.0 + viewport->translate[2];
-      range_max = viewport->scale[2] *  1.0 + viewport->translate[2];
 
-      /* D3D (and by implication SVGA) doesn't like dealing with zmax
-       * less than zmin.  Detect that case, flip the depth range and
-       * invert our z-scale factor to achieve the same effect.
-       */
-      if (range_min > range_max) {
-         float range_tmp;
-         range_tmp = range_min; 
-         range_min = range_max; 
-         range_max = range_tmp;
-         prescale.scale[2]     = -prescale.scale[2];
-      }
+   range_min = viewport->scale[2] * -1.0 + viewport->translate[2];
+   range_max = viewport->scale[2] *  1.0 + viewport->translate[2];
+
+   /* D3D (and by implication SVGA) doesn't like dealing with zmax
+    * less than zmin.  Detect that case, flip the depth range and
+    * invert our z-scale factor to achieve the same effect.
+    */
+   if (range_min > range_max) {
+      float range_tmp;
+      range_tmp = range_min; 
+      range_min = range_max; 
+      range_max = range_tmp;
+      prescale.scale[2]     = -prescale.scale[2];
    }
 
    if (prescale.enabled) {
diff --git a/src/gallium/drivers/svga/svga_state_need_swtnl.c b/src/gallium/drivers/svga/svga_state_need_swtnl.c
index dd13a89d24..a6ed18599c 100644
--- a/src/gallium/drivers/svga/svga_state_need_swtnl.c
+++ b/src/gallium/drivers/svga/svga_state_need_swtnl.c
@@ -129,8 +129,7 @@ static int update_need_pipeline( struct svga_context *svga,
 
    /* SVGA_NEW_CLIP 
     */
-   if (!svga->curr.rast->templ.bypass_vs_clip_and_viewport &&
-       svga->curr.clip.nr) {
+   if (svga->curr.clip.nr) {
       SVGA_DBG(DEBUG_SWTNL, "%s: userclip\n", __FUNCTION__);
       need_pipeline = TRUE;
    }
diff --git a/src/gallium/drivers/trace/tr_dump_state.c b/src/gallium/drivers/trace/tr_dump_state.c
index 6da186a655..f97d963dba 100644
--- a/src/gallium/drivers/trace/tr_dump_state.c
+++ b/src/gallium/drivers/trace/tr_dump_state.c
@@ -112,7 +112,6 @@ void trace_dump_rasterizer_state(const struct pipe_rasterizer_state *state)
    trace_dump_member(uint, state, line_stipple_factor);
    trace_dump_member(uint, state, line_stipple_pattern);
    trace_dump_member(bool, state, line_last_pixel);
-   trace_dump_member(bool, state, bypass_vs_clip_and_viewport);
    trace_dump_member(bool, state, flatshade_first);
    trace_dump_member(bool, state, gl_rasterization_rules);