Merge branch 'master' of git+ssh://michal@git.freedesktop.org/git/mesa/mesa into softpipe_0_1_branch

20 files changed, 493 insertions, 423 deletions

diff --git a/src/mesa/drivers/dri/i810/i810state.c b/src/mesa/drivers/dri/i810/i810state.c
index 3ad25282d9..e0d5b2b487 100644
--- a/src/mesa/drivers/dri/i810/i810state.c
+++ b/src/mesa/drivers/dri/i810/i810state.c
@@ -380,7 +380,10 @@ static void i810CullFaceFrontFace(GLcontext *ctx, GLenum unused)
 static void i810LineWidth( GLcontext *ctx, GLfloat widthf )
 {
    i810ContextPtr imesa = I810_CONTEXT( ctx );
-   int width = (int)ctx->Line._Width;
+   /* AA, non-AA limits are same */
+   const int width = (int) CLAMP(ctx->Line.Width,
+                                 ctx->Const.MinLineWidth,
+                                 ctx->Const.MaxLineWidth);
 
    imesa->LcsLineWidth = 0;
    if (width & 1) imesa->LcsLineWidth |= LCS_LINEWIDTH_1_0;
@@ -396,7 +399,10 @@ static void i810LineWidth( GLcontext *ctx, GLfloat widthf )
 static void i810PointSize( GLcontext *ctx, GLfloat sz )
 {
    i810ContextPtr imesa = I810_CONTEXT( ctx );
-   int size = (int)ctx->Point._Size;
+   /* AA, non-AA limits are same */
+   const int size = (int) CLAMP(ctx->Point.Size,
+                                ctx->Const.MinPointSize,
+                                ctx->Const.MaxPointSize);
 
    imesa->LcsPointSize = 0;
    if (size & 1) imesa->LcsPointSize |= LCS_LINEWIDTH_1_0;
diff --git a/src/mesa/drivers/dri/i810/i810tris.c b/src/mesa/drivers/dri/i810/i810tris.c
index 3e09427bb9..2c4ee06633 100644
--- a/src/mesa/drivers/dri/i810/i810tris.c
+++ b/src/mesa/drivers/dri/i810/i810tris.c
@@ -112,7 +112,9 @@ static __inline__ void i810_draw_quad( i810ContextPtr imesa,
 static __inline__ void i810_draw_point( i810ContextPtr imesa,
 					i810VertexPtr tmp )
 {
-   GLfloat sz = imesa->glCtx->Point._Size * .5;
+   GLfloat sz = 0.5 * CLAMP(imesa->glCtx->Point.Size,
+                            imesa->glCtx->Const.MinPointSize,
+                            imesa->glCtx->Const.MaxPointSize);
    int vertsize = imesa->vertex_size;
    GLuint *vb = i810AllocDmaLow( imesa, 2 * 4 * vertsize );
    int j;
diff --git a/src/mesa/drivers/dri/i915tex/intel_batchbuffer.h b/src/mesa/drivers/dri/i915tex/intel_batchbuffer.h
index 59261f7274..212f130101 100644
--- a/src/mesa/drivers/dri/i915tex/intel_batchbuffer.h
+++ b/src/mesa/drivers/dri/i915tex/intel_batchbuffer.h
@@ -9,7 +9,7 @@ struct intel_context;
 #define BATCH_SZ 16384
 #define BATCH_RESERVED 16
 
-#define MAX_RELOCS 400
+#define MAX_RELOCS 4096
 
 #define INTEL_BATCH_NO_CLIPRECTS 0x1
 #define INTEL_BATCH_CLIPRECTS    0x2
diff --git a/src/mesa/drivers/dri/i915tex/intel_fbo.c b/src/mesa/drivers/dri/i915tex/intel_fbo.c
index 2dd3617f93..a09db46163 100644
--- a/src/mesa/drivers/dri/i915tex/intel_fbo.c
+++ b/src/mesa/drivers/dri/i915tex/intel_fbo.c
@@ -509,7 +509,7 @@ intel_new_renderbuffer(GLcontext * ctx, GLuint name)
  */
 static void
 intel_bind_framebuffer(GLcontext * ctx, GLenum target,
-                       struct gl_framebuffer *fb)
+                       struct gl_framebuffer *fb, struct gl_framebuffer *fbread)
 {
    if (target == GL_FRAMEBUFFER_EXT || target == GL_DRAW_FRAMEBUFFER_EXT) {
       intel_draw_buffer(ctx, fb);
diff --git a/src/mesa/drivers/dri/i965/brw_sf_state.c b/src/mesa/drivers/dri/i965/brw_sf_state.c
index bfac52d765..9a6e5f5f19 100644
--- a/src/mesa/drivers/dri/i965/brw_sf_state.c
+++ b/src/mesa/drivers/dri/i965/brw_sf_state.c
@@ -173,7 +173,8 @@ static void upload_sf_unit( struct brw_context *brw )
       
 
    /* _NEW_LINE */
-   sf.sf6.line_width = brw->attribs.Line->_Width * (1<<1);
+   /* XXX use ctx->Const.Min/MaxLineWidth here */
+   sf.sf6.line_width = CLAMP(brw->attribs.Line->Width, 1.0, 5.0) * (1<<1);
 
    sf.sf6.line_endcap_aa_region_width = 1;
    if (brw->attribs.Line->SmoothFlag)
@@ -183,7 +184,8 @@ static void upload_sf_unit( struct brw_context *brw )
 
    /* _NEW_POINT */
    sf.sf6.point_rast_rule = 1;	/* opengl conventions */
-   sf.sf7.point_size = brw->attribs.Point->_Size * (1<<3);
+   /* XXX clamp max depends on AA vs. non-AA */
+   sf.sf7.point_size = CLAMP(brw->attribs.Point->Size, 1.0, 3.0) * (1<<3);
    sf.sf7.use_point_size_state = !brw->attribs.Point->_Attenuated;
       
    /* might be BRW_NEW_PRIMITIVE if we have to adjust pv for polygons:
diff --git a/src/mesa/drivers/dri/mach64/mach64_native_vb.c b/src/mesa/drivers/dri/mach64/mach64_native_vb.c
index 75cf0e2ed2..248fa2a9a2 100644
--- a/src/mesa/drivers/dri/mach64/mach64_native_vb.c
+++ b/src/mesa/drivers/dri/mach64/mach64_native_vb.c
@@ -103,7 +103,7 @@ void TAG(translate_vertex)(GLcontext *ctx,
 
    assert( p + 1 - (CARD32 *)src == 10 );
 	 
-   dst->pointSize = ctx->Point._Size;
+   dst->pointSize = ctx->Point.Size;
 }
 
 
diff --git a/src/mesa/drivers/dri/mach64/mach64_tris.c b/src/mesa/drivers/dri/mach64/mach64_tris.c
index 08cc1849a1..369f610442 100644
--- a/src/mesa/drivers/dri/mach64/mach64_tris.c
+++ b/src/mesa/drivers/dri/mach64/mach64_tris.c
@@ -673,7 +673,10 @@ static __inline void mach64_draw_line( mach64ContextPtr mmesa,
 #if MACH64_NATIVE_VTXFMT
    GLcontext *ctx = mmesa->glCtx;
    const GLuint vertsize = mmesa->vertex_size;
-   GLint width = (GLint)(mmesa->glCtx->Line._Width * 2.0); /* 2 fractional bits for hardware */
+   /* 2 fractional bits for hardware: */
+   const int width = (int) (2.0 * CLAMP(mmesa->glCtx->Line.Width,
+                                        mmesa->glCtx->Const.MinLineWidth,
+                                        mmesa->glCtx->Const.MaxLineWidth));
    GLfloat ooa;
    GLuint *pxy0, *pxy1;
    GLuint xy0old, xy0, xy1old, xy1;
@@ -691,9 +694,6 @@ static __inline void mach64_draw_line( mach64ContextPtr mmesa,
       mach64_print_vertex( ctx, v1 );
    }
   
-   if( !width )
-      width = 1;	/* round to the nearest supported width */
-      
    pxy0 = &v0->ui[xyoffset];
    xy0old = *pxy0;
    xy0 = LE32_IN( &xy0old );
@@ -961,7 +961,10 @@ static __inline void mach64_draw_point( mach64ContextPtr mmesa,
 #if MACH64_NATIVE_VTXFMT
    GLcontext *ctx = mmesa->glCtx;
    const GLuint vertsize = mmesa->vertex_size;
-   GLint sz = (GLint)(mmesa->glCtx->Point._Size * 2.0); /* 2 fractional bits for hardware */
+   /* 2 fractional bits for hardware: */
+   GLint sz = (GLint) (2.0 * CLAMP(mmesa->glCtx->Point.Size,
+                                   ctx->Const.MinPointSize,
+                                   ctx->Const.MaxPointSize));
    GLfloat ooa;
    GLuint *pxy;
    GLuint xyold, xy;
diff --git a/src/mesa/drivers/dri/mga/mgatris.c b/src/mesa/drivers/dri/mga/mgatris.c
index 2b7ea05b14..91b413ae76 100644
--- a/src/mesa/drivers/dri/mga/mgatris.c
+++ b/src/mesa/drivers/dri/mga/mgatris.c
@@ -104,8 +104,10 @@ static void __inline__ mga_draw_quad( mgaContextPtr mmesa,
 static __inline__ void mga_draw_point( mgaContextPtr mmesa,
 					mgaVertexPtr tmp )
 {
-   GLfloat sz = mmesa->glCtx->Point._Size * .5;
-   int vertex_size = mmesa->vertex_size;
+   const GLfloat sz = 0.5 * CLAMP(mmesa->glCtx->Point.Size,
+                                  mmesa->glCtx->Const.MinPointSize,
+                                  mmesa->glCtx->Const.MaxPointSize);
+   const int vertex_size = mmesa->vertex_size;
    GLuint *vb = mgaAllocDmaLow( mmesa, 6 * 4 * vertex_size );
    int j;
    
@@ -165,7 +167,9 @@ static __inline__ void mga_draw_line( mgaContextPtr mmesa,
    GLuint vertex_size = mmesa->vertex_size;
    GLuint *vb = mgaAllocDmaLow( mmesa, 6 * 4 * vertex_size );
    GLfloat dx, dy, ix, iy;
-   GLfloat width = mmesa->glCtx->Line._Width;
+   const GLfloat width = CLAMP(mmesa->glCtx->Line.Width,
+                               mmesa->glCtx->Const.MinLineWidth,
+                               mmesa->glCtx->Const.MaxLineWidth);
    GLint j;
 
 #if 0
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_buffers.c b/src/mesa/drivers/dri/nouveau/nouveau_buffers.c
index f98d666563..d498f616c9 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_buffers.c
+++ b/src/mesa/drivers/dri/nouveau/nouveau_buffers.c
@@ -390,9 +390,12 @@ nouveauNewRenderbuffer(GLcontext *ctx, GLuint name)
 }
 
 static void
-nouveauBindFramebuffer(GLcontext *ctx, GLenum target, struct gl_framebuffer *fb)
+nouveauBindFramebuffer(GLcontext *ctx, GLenum target,
+                       struct gl_framebuffer *fb, struct gl_framebuffer *fbread)
 {
-   nouveau_build_framebuffer(ctx, fb);
+   if (target == GL_FRAMEBUFFER_EXT || target == GL_DRAW_FRAMEBUFFER_EXT) {
+      nouveau_build_framebuffer(ctx, fb);
+   }
 }
 
 static void
diff --git a/src/mesa/drivers/dri/r200/r200_state.c b/src/mesa/drivers/dri/r200/r200_state.c
index 2115799b9b..1d975ecd57 100644
--- a/src/mesa/drivers/dri/r200/r200_state.c
+++ b/src/mesa/drivers/dri/r200/r200_state.c
@@ -772,9 +772,11 @@ static void r200LineWidth( GLcontext *ctx, GLfloat widthf )
    R200_STATECHANGE( rmesa, set );
 
    /* Line width is stored in U6.4 format.
+    * Same min/max limits for AA, non-AA lines.
     */
    rmesa->hw.lin.cmd[LIN_SE_LINE_WIDTH] &= ~0xffff;
-   rmesa->hw.lin.cmd[LIN_SE_LINE_WIDTH] |= (GLuint)(ctx->Line._Width * 16.0);
+   rmesa->hw.lin.cmd[LIN_SE_LINE_WIDTH] |= (GLuint)
+      (CLAMP(widthf, ctx->Const.MinLineWidth, ctx->Const.MaxLineWidth) * 16.0);
 
    if ( widthf > 1.0 ) {
       rmesa->hw.set.cmd[SET_SE_CNTL] |=  R200_WIDELINE_ENABLE;
diff --git a/src/mesa/drivers/dri/r300/r300_context.h b/src/mesa/drivers/dri/r300/r300_context.h
index 0349bac9a2..be6909724a 100644
--- a/src/mesa/drivers/dri/r300/r300_context.h
+++ b/src/mesa/drivers/dri/r300/r300_context.h
@@ -568,7 +568,7 @@ struct r300_vertex_shader_fragment {
 	union {
 		GLuint d[VSF_MAX_FRAGMENT_LENGTH];
 		float f[VSF_MAX_FRAGMENT_LENGTH];
-		struct r300_vertprog_instruction i[VSF_MAX_FRAGMENT_LENGTH / 4];
+		GLuint i[VSF_MAX_FRAGMENT_LENGTH];
 	} body;
 };
 
diff --git a/src/mesa/drivers/dri/r300/r300_state.c b/src/mesa/drivers/dri/r300/r300_state.c
index 6789efd428..088216c76e 100644
--- a/src/mesa/drivers/dri/r300/r300_state.c
+++ b/src/mesa/drivers/dri/r300/r300_state.c
@@ -733,8 +733,8 @@ static void r300Fogfv(GLcontext * ctx, GLenum pname, const GLfloat * param)
 static void r300PointSize(GLcontext * ctx, GLfloat size)
 {
 	r300ContextPtr r300 = R300_CONTEXT(ctx);
-
-	size = ctx->Point._Size;
+        /* same size limits for AA, non-AA points */
+	size = CLAMP(size, ctx->Const.MinPointSize, ctx->Const.MaxPointSize);
 
 	R300_STATECHANGE(r300, ps);
 	r300->hw.ps.cmd[R300_PS_POINTSIZE] =
@@ -749,8 +749,9 @@ static void r300LineWidth(GLcontext * ctx, GLfloat widthf)
 {
 	r300ContextPtr r300 = R300_CONTEXT(ctx);
 
-	widthf = ctx->Line._Width;
-
+	widthf = CLAMP(widthf,
+                       ctx->Const.MinPointSize,
+                       ctx->Const.MaxPointSize);
 	R300_STATECHANGE(r300, lcntl);
 	r300->hw.lcntl.cmd[1] =
 	    R300_LINE_CNT_HO | R300_LINE_CNT_VE | (int)(widthf * 6.0);
@@ -1566,30 +1567,31 @@ static void r300SetupDefaultVertexProgram(r300ContextPtr rmesa)
 
 	for (i = VERT_ATTRIB_POS; i < VERT_ATTRIB_MAX; i++) {
 		if (rmesa->state.sw_tcl_inputs[i] != -1) {
-			prog->program.body.i[program_end].opcode = EASY_VSF_OP(MUL, o_reg++, ALL, RESULT);
-			prog->program.body.i[program_end].src[0] = VSF_REG(rmesa->state.sw_tcl_inputs[i]);
-			prog->program.body.i[program_end].src[1] = VSF_ATTR_UNITY(rmesa->state.sw_tcl_inputs[i]);
-			prog->program.body.i[program_end].src[2] = VSF_UNITY(rmesa->state.sw_tcl_inputs[i]);
-			program_end++;
+			prog->program.body.i[program_end + 0] = EASY_VSF_OP(MUL, o_reg++, ALL, RESULT);
+			prog->program.body.i[program_end + 1] = VSF_REG(rmesa->state.sw_tcl_inputs[i]);
+			prog->program.body.i[program_end + 2] = VSF_ATTR_UNITY(rmesa->state.sw_tcl_inputs[i]);
+			prog->program.body.i[program_end + 3] = VSF_UNITY(rmesa->state.sw_tcl_inputs[i]);
+			program_end += 4;
 		}
 	}
 
-	prog->program.length = program_end * 4;
+	prog->program.length = program_end;
 
-	r300SetupVertexProgramFragment(rmesa, R300_PVS_UPLOAD_PROGRAM, &(prog->program));
+	r300SetupVertexProgramFragment(rmesa, R300_PVS_UPLOAD_PROGRAM,
+				       &(prog->program));
 	inst_count = (prog->program.length / 4) - 1;
 
 	R300_STATECHANGE(rmesa, pvs);
 	rmesa->hw.pvs.cmd[R300_PVS_CNTL_1] =
-	  (0 << R300_PVS_CNTL_1_PROGRAM_START_SHIFT) |
-	  (inst_count << R300_PVS_CNTL_1_POS_END_SHIFT) |
-	  (inst_count << R300_PVS_CNTL_1_PROGRAM_END_SHIFT);
+	    (0 << R300_PVS_CNTL_1_PROGRAM_START_SHIFT) |
+	    (inst_count << R300_PVS_CNTL_1_POS_END_SHIFT) |
+	    (inst_count << R300_PVS_CNTL_1_PROGRAM_END_SHIFT);
 	rmesa->hw.pvs.cmd[R300_PVS_CNTL_2] =
-	  (0 << R300_PVS_CNTL_2_PARAM_OFFSET_SHIFT) |
-	  (param_count << R300_PVS_CNTL_2_PARAM_COUNT_SHIFT);
+	    (0 << R300_PVS_CNTL_2_PARAM_OFFSET_SHIFT) |
+	    (param_count << R300_PVS_CNTL_2_PARAM_COUNT_SHIFT);
 	rmesa->hw.pvs.cmd[R300_PVS_CNTL_3] =
-	  (inst_count << R300_PVS_CNTL_3_PROGRAM_UNKNOWN_SHIFT) |
-	  (inst_count << R300_PVS_CNTL_3_PROGRAM_UNKNOWN2_SHIFT);
+	    (inst_count << R300_PVS_CNTL_3_PROGRAM_UNKNOWN_SHIFT) |
+	    (inst_count << R300_PVS_CNTL_3_PROGRAM_UNKNOWN2_SHIFT);
 }
 
 static void r300SetupRealVertexProgram(r300ContextPtr rmesa)
diff --git a/src/mesa/drivers/dri/r300/r300_vertprog.c b/src/mesa/drivers/dri/r300/r300_vertprog.c
index 0fb6110494..7d4e8c9511 100644
--- a/src/mesa/drivers/dri/r300/r300_vertprog.c
+++ b/src/mesa/drivers/dri/r300/r300_vertprog.c
@@ -31,6 +31,8 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
  * \author Aapo Tahkola <aet@rasterburn.org>
  *
  * \author Oliver McFadden <z3ro.geek@gmail.com>
+ *
+ * For a description of the vertex program instruction set see r300_reg.h.
  */
 
 #include "glheader.h"
@@ -366,20 +368,19 @@ static GLboolean valid_dst(struct r300_vertex_program *vp,
  * component.
  */
 
-static void t_opcode_abs(struct r300_vertex_program *vp,
-			 struct prog_instruction *vpi,
-			 struct r300_vertprog_instruction *o_inst,
-			 struct prog_src_register src[3])
+static GLuint *t_opcode_abs(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
 {
 	//MAX RESULT 1.X Y Z W PARAM 0{} {X Y Z W} PARAM 0{X Y Z W } {X Y Z W} neg Xneg Yneg Zneg W
 
-	o_inst->opcode =
+	inst[0] =
 	    MAKE_VSF_OP(R300_VPI_OUT_OP_MAX, t_dst_index(vp, &vpi->DstReg),
 			t_dst_mask(vpi->DstReg.WriteMask),
 			t_dst_class(vpi->DstReg.File));
 
-	o_inst->src[0] = t_src(vp, &src[0]);
-	o_inst->src[1] =
+	inst[1] = t_src(vp, &src[0]);
+	inst[2] =
 	    MAKE_VSF_SOURCE(t_src_index(vp, &src[0]),
 			    t_swizzle(GET_SWZ(src[0].Swizzle, 0)),
 			    t_swizzle(GET_SWZ(src[0].Swizzle, 1)),
@@ -389,13 +390,14 @@ static void t_opcode_abs(struct r300_vertex_program *vp,
 			    (!src[0].
 			     NegateBase) ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
 	    (src[0].RelAddr << 4);
-	o_inst->src[2] = 0;
+	inst[3] = 0;
+
+	return inst;
 }
 
-static void t_opcode_add(struct r300_vertex_program *vp,
-			 struct prog_instruction *vpi,
-			 struct r300_vertprog_instruction *o_inst,
-			 struct prog_src_register src[3])
+static GLuint *t_opcode_add(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
 {
 	unsigned long hw_op;
 
@@ -405,53 +407,55 @@ static void t_opcode_add(struct r300_vertex_program *vp,
 		 PROGRAM_TEMPORARY) ? R300_VPI_OUT_OP_MAD_2 :
 	    R300_VPI_OUT_OP_MAD;
 
-	o_inst->opcode =
+	inst[0] =
 	    MAKE_VSF_OP(hw_op, t_dst_index(vp, &vpi->DstReg),
 			t_dst_mask(vpi->DstReg.WriteMask),
 			t_dst_class(vpi->DstReg.File));
-	o_inst->src[0] = ONE_SRC_0;
-	o_inst->src[1] = t_src(vp, &src[0]);
-	o_inst->src[2] = t_src(vp, &src[1]);
+	inst[1] = ONE_SRC_0;
+	inst[2] = t_src(vp, &src[0]);
+	inst[3] = t_src(vp, &src[1]);
 #else
-	o_inst->opcode =
+	inst[0] =
 	    MAKE_VSF_OP(R300_VPI_OUT_OP_ADD, t_dst_index(vp, &vpi->DstReg),
 			t_dst_mask(vpi->DstReg.WriteMask),
 			t_dst_class(vpi->DstReg.File));
-	o_inst->src[0] = t_src(vp, &src[0]);
-	o_inst->src[1] = t_src(vp, &src[1]);
-	o_inst->src[2] = ZERO_SRC_1;
+	inst[1] = t_src(vp, &src[0]);
+	inst[2] = t_src(vp, &src[1]);
+	inst[3] = ZERO_SRC_1;
 
 #endif
+
+	return inst;
 }
 
-static void t_opcode_arl(struct r300_vertex_program *vp,
-			 struct prog_instruction *vpi,
-			 struct r300_vertprog_instruction *o_inst,
-			 struct prog_src_register src[3])
+static GLuint *t_opcode_arl(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
 {
-	o_inst->opcode =
+	inst[0] =
 	    MAKE_VSF_OP(R300_VPI_OUT_OP_ARL, t_dst_index(vp, &vpi->DstReg),
 			t_dst_mask(vpi->DstReg.WriteMask),
 			t_dst_class(vpi->DstReg.File));
 
-	o_inst->src[0] = t_src(vp, &src[0]);
-	o_inst->src[1] = ZERO_SRC_0;
-	o_inst->src[2] = ZERO_SRC_0;
+	inst[1] = t_src(vp, &src[0]);
+	inst[2] = ZERO_SRC_0;
+	inst[3] = ZERO_SRC_0;
+
+	return inst;
 }
 
-static void t_opcode_dp3(struct r300_vertex_program *vp,
-			 struct prog_instruction *vpi,
-			 struct r300_vertprog_instruction *o_inst,
-			 struct prog_src_register src[3])
+static GLuint *t_opcode_dp3(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
 {
 	//DOT RESULT 1.X Y Z W PARAM 0{} {X Y Z ZERO} PARAM 0{} {X Y Z ZERO}
 
-	o_inst->opcode =
+	inst[0] =
 	    MAKE_VSF_OP(R300_VPI_OUT_OP_DOT, t_dst_index(vp, &vpi->DstReg),
 			t_dst_mask(vpi->DstReg.WriteMask),
 			t_dst_class(vpi->DstReg.File));
 
-	o_inst->src[0] =
+	inst[1] =
 	    MAKE_VSF_SOURCE(t_src_index(vp, &src[0]),
 			    t_swizzle(GET_SWZ(src[0].Swizzle, 0)),
 			    t_swizzle(GET_SWZ(src[0].Swizzle, 1)),
@@ -461,7 +465,7 @@ static void t_opcode_dp3(struct r300_vertex_program *vp,
 			    NegateBase ? VSF_FLAG_XYZ : VSF_FLAG_NONE) |
 	    (src[0].RelAddr << 4);
 
-	o_inst->src[1] =
+	inst[2] =
 	    MAKE_VSF_SOURCE(t_src_index(vp, &src[1]),
 			    t_swizzle(GET_SWZ(src[1].Swizzle, 0)),
 			    t_swizzle(GET_SWZ(src[1].Swizzle, 1)),
@@ -471,36 +475,38 @@ static void t_opcode_dp3(struct r300_vertex_program *vp,
 			    NegateBase ? VSF_FLAG_XYZ : VSF_FLAG_NONE) |
 	    (src[1].RelAddr << 4);
 
-	o_inst->src[2] = ZERO_SRC_1;
+	inst[3] = ZERO_SRC_1;
+
+	return inst;
 }
 
-static void t_opcode_dp4(struct r300_vertex_program *vp,
-			 struct prog_instruction *vpi,
-			 struct r300_vertprog_instruction *o_inst,
-			 struct prog_src_register src[3])
+static GLuint *t_opcode_dp4(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
 {
-	o_inst->opcode =
+	inst[0] =
 	    MAKE_VSF_OP(R300_VPI_OUT_OP_DOT, t_dst_index(vp, &vpi->DstReg),
 			t_dst_mask(vpi->DstReg.WriteMask),
 			t_dst_class(vpi->DstReg.File));
 
-	o_inst->src[0] = t_src(vp, &src[0]);
-	o_inst->src[1] = t_src(vp, &src[1]);
-	o_inst->src[2] = ZERO_SRC_1;
+	inst[1] = t_src(vp, &src[0]);
+	inst[2] = t_src(vp, &src[1]);
+	inst[3] = ZERO_SRC_1;
+
+	return inst;
 }
 
-static void t_opcode_dph(struct r300_vertex_program *vp,
-			 struct prog_instruction *vpi,
-			 struct r300_vertprog_instruction *o_inst,
-			 struct prog_src_register src[3])
+static GLuint *t_opcode_dph(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
 {
 	//DOT RESULT 1.X Y Z W PARAM 0{} {X Y Z ONE} PARAM 0{} {X Y Z W}
-	o_inst->opcode =
+	inst[0] =
 	    MAKE_VSF_OP(R300_VPI_OUT_OP_DOT, t_dst_index(vp, &vpi->DstReg),
 			t_dst_mask(vpi->DstReg.WriteMask),
 			t_dst_class(vpi->DstReg.File));
 
-	o_inst->src[0] =
+	inst[1] =
 	    MAKE_VSF_SOURCE(t_src_index(vp, &src[0]),
 			    t_swizzle(GET_SWZ(src[0].Swizzle, 0)),
 			    t_swizzle(GET_SWZ(src[0].Swizzle, 1)),
@@ -509,81 +515,84 @@ static void t_opcode_dph(struct r300_vertex_program *vp,
 			    src[0].
 			    NegateBase ? VSF_FLAG_XYZ : VSF_FLAG_NONE) |
 	    (src[0].RelAddr << 4);
-	o_inst->src[1] = t_src(vp, &src[1]);
-	o_inst->src[2] = ZERO_SRC_1;
+	inst[2] = t_src(vp, &src[1]);
+	inst[3] = ZERO_SRC_1;
+
+	return inst;
 }
 
-static void t_opcode_dst(struct r300_vertex_program *vp,
-			 struct prog_instruction *vpi,
-			 struct r300_vertprog_instruction *o_inst,
-			 struct prog_src_register src[3])
+static GLuint *t_opcode_dst(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
 {
-	o_inst->opcode =
+	inst[0] =
 	    MAKE_VSF_OP(R300_VPI_OUT_OP_DST, t_dst_index(vp, &vpi->DstReg),
 			t_dst_mask(vpi->DstReg.WriteMask),
 			t_dst_class(vpi->DstReg.File));
 
-	o_inst->src[0] = t_src(vp, &src[0]);
-	o_inst->src[1] = t_src(vp, &src[1]);
-	o_inst->src[2] = ZERO_SRC_1;
+	inst[1] = t_src(vp, &src[0]);
+	inst[2] = t_src(vp, &src[1]);
+	inst[3] = ZERO_SRC_1;
+
+	return inst;
 }
 
-static void t_opcode_ex2(struct r300_vertex_program *vp,
-			 struct prog_instruction *vpi,
-			 struct r300_vertprog_instruction *o_inst,
-			 struct prog_src_register src[3])
+static GLuint *t_opcode_ex2(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
 {
-	o_inst->opcode =
+	inst[0] =
 	    MAKE_VSF_OP(R300_VPI_OUT_OP_EX2, t_dst_index(vp, &vpi->DstReg),
 			t_dst_mask(vpi->DstReg.WriteMask),
 			t_dst_class(vpi->DstReg.File));
 
-	o_inst->src[0] = t_src_scalar(vp, &src[0]);
-	o_inst->src[1] = ZERO_SRC_0;
-	o_inst->src[2] = ZERO_SRC_0;
+	inst[1] = t_src_scalar(vp, &src[0]);
+	inst[2] = ZERO_SRC_0;
+	inst[3] = ZERO_SRC_0;
+
+	return inst;
 }
 
-static void t_opcode_exp(struct r300_vertex_program *vp,
-			 struct prog_instruction *vpi,
-			 struct r300_vertprog_instruction *o_inst,
-			 struct prog_src_register src[3])
+static GLuint *t_opcode_exp(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
 {
-	o_inst->opcode =
+	inst[0] =
 	    MAKE_VSF_OP(R300_VPI_OUT_OP_EXP, t_dst_index(vp, &vpi->DstReg),
 			t_dst_mask(vpi->DstReg.WriteMask),
 			t_dst_class(vpi->DstReg.File));
 
-	o_inst->src[0] = t_src_scalar(vp, &src[0]);
-	o_inst->src[1] = ZERO_SRC_0;
-	o_inst->src[2] = ZERO_SRC_0;
+	inst[1] = t_src_scalar(vp, &src[0]);
+	inst[2] = ZERO_SRC_0;
+	inst[3] = ZERO_SRC_0;
+
+	return inst;
 }
 
-static struct r300_vertprog_instruction *t_opcode_flr(struct r300_vertex_program *vp,
-						      struct prog_instruction *vpi,
-						      struct r300_vertprog_instruction *o_inst,
-						      struct prog_src_register src[3],
-						      int *u_temp_i)
+static GLuint *t_opcode_flr(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3], int *u_temp_i)
 {
 	/* FRC TMP 0.X Y Z W PARAM 0{} {X Y Z W}
 	   ADD RESULT 1.X Y Z W PARAM 0{} {X Y Z W} TMP 0{X Y Z W } {X Y Z W} neg Xneg Yneg Zneg W */
 
-	o_inst->opcode =
+	inst[0] =
 	    MAKE_VSF_OP(R300_VPI_OUT_OP_FRC, *u_temp_i,
 			t_dst_mask(vpi->DstReg.WriteMask),
 			VSF_OUT_CLASS_TMP);
 
-	o_inst->src[0] = t_src(vp, &src[0]);
-	o_inst->src[1] = ZERO_SRC_0;
-	o_inst->src[2] = ZERO_SRC_0;
-	o_inst++;
+	inst[1] = t_src(vp, &src[0]);
+	inst[2] = ZERO_SRC_0;
+	inst[3] = ZERO_SRC_0;
+	inst += 4;
 
-	o_inst->opcode =
+	inst[0] =
 	    MAKE_VSF_OP(R300_VPI_OUT_OP_ADD, t_dst_index(vp, &vpi->DstReg),
 			t_dst_mask(vpi->DstReg.WriteMask),
 			t_dst_class(vpi->DstReg.File));
 
-	o_inst->src[0] = t_src(vp, &src[0]);
-	o_inst->src[1] =
+	inst[1] = t_src(vp, &src[0]);
+	inst[2] =
 	    MAKE_VSF_SOURCE(*u_temp_i, VSF_IN_COMPONENT_X,
 			    VSF_IN_COMPONENT_Y, VSF_IN_COMPONENT_Z,
 			    VSF_IN_COMPONENT_W, VSF_IN_CLASS_TMP,
@@ -592,40 +601,40 @@ static struct r300_vertprog_instruction *t_opcode_flr(struct r300_vertex_program
 			     NegateBase) ? VSF_FLAG_ALL : VSF_FLAG_NONE
 			    /*VSF_FLAG_ALL */ );
 
-	o_inst->src[2] = ZERO_SRC_0;
+	inst[3] = ZERO_SRC_0;
 	(*u_temp_i)--;
 
-	return o_inst;
+	return inst;
 }
 
-static void t_opcode_frc(struct r300_vertex_program *vp,
-			 struct prog_instruction *vpi,
-			 struct r300_vertprog_instruction *o_inst,
-			 struct prog_src_register src[3])
+static GLuint *t_opcode_frc(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
 {
-	o_inst->opcode =
+	inst[0] =
 	    MAKE_VSF_OP(R300_VPI_OUT_OP_FRC, t_dst_index(vp, &vpi->DstReg),
 			t_dst_mask(vpi->DstReg.WriteMask),
 			t_dst_class(vpi->DstReg.File));
 
-	o_inst->src[0] = t_src(vp, &src[0]);
-	o_inst->src[1] = ZERO_SRC_0;
-	o_inst->src[2] = ZERO_SRC_0;
+	inst[1] = t_src(vp, &src[0]);
+	inst[2] = ZERO_SRC_0;
+	inst[3] = ZERO_SRC_0;
+
+	return inst;
 }
 
-static void t_opcode_lg2(struct r300_vertex_program *vp,
-			 struct prog_instruction *vpi,
-			 struct r300_vertprog_instruction *o_inst,
-			 struct prog_src_register src[3])
+static GLuint *t_opcode_lg2(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
 {
 	// LG2 RESULT 1.X Y Z W PARAM 0{} {X X X X}
 
-	o_inst->opcode =
+	inst[0] =
 	    MAKE_VSF_OP(R300_VPI_OUT_OP_LG2, t_dst_index(vp, &vpi->DstReg),
 			t_dst_mask(vpi->DstReg.WriteMask),
 			t_dst_class(vpi->DstReg.File));
 
-	o_inst->src[0] =
+	inst[1] =
 	    MAKE_VSF_SOURCE(t_src_index(vp, &src[0]),
 			    t_swizzle(GET_SWZ(src[0].Swizzle, 0)),
 			    t_swizzle(GET_SWZ(src[0].Swizzle, 0)),
@@ -635,70 +644,70 @@ static void t_opcode_lg2(struct r300_vertex_program *vp,
 			    src[0].
 			    NegateBase ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
 	    (src[0].RelAddr << 4);
-	o_inst->src[1] = ZERO_SRC_0;
-	o_inst->src[2] = ZERO_SRC_0;
+	inst[2] = ZERO_SRC_0;
+	inst[3] = ZERO_SRC_0;
+
+	return inst;
 }
 
-static void t_opcode_lit(struct r300_vertex_program *vp,
-			 struct prog_instruction *vpi,
-			 struct r300_vertprog_instruction *o_inst,
-			 struct prog_src_register src[3])
+static GLuint *t_opcode_lit(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
 {
 	//LIT TMP 1.Y Z TMP 1{} {X W Z Y} TMP 1{} {Y W Z X} TMP 1{} {Y X Z W}
 
-	o_inst->opcode =
+	inst[0] =
 	    MAKE_VSF_OP(R300_VPI_OUT_OP_LIT, t_dst_index(vp, &vpi->DstReg),
 			t_dst_mask(vpi->DstReg.WriteMask),
 			t_dst_class(vpi->DstReg.File));
 	/* NOTE: Users swizzling might not work. */
-	o_inst->src[0] = MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), t_swizzle(GET_SWZ(src[0].Swizzle, 0)),	// x
-					 t_swizzle(GET_SWZ(src[0].Swizzle, 3)),	// w
-					 VSF_IN_COMPONENT_ZERO,	// z
-					 t_swizzle(GET_SWZ(src[0].Swizzle, 1)),	// y
-					 t_src_class(src[0].File),
-					 src[0].
-					 NegateBase ? VSF_FLAG_ALL :
-					 VSF_FLAG_NONE) | (src[0].
-							   RelAddr << 4);
-	o_inst->src[1] = MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), t_swizzle(GET_SWZ(src[0].Swizzle, 1)),	// y
-					 t_swizzle(GET_SWZ(src[0].Swizzle, 3)),	// w
-					 VSF_IN_COMPONENT_ZERO,	// z
-					 t_swizzle(GET_SWZ(src[0].Swizzle, 0)),	// x
-					 t_src_class(src[0].File),
-					 src[0].
-					 NegateBase ? VSF_FLAG_ALL :
-					 VSF_FLAG_NONE) | (src[0].
-							   RelAddr << 4);
-	o_inst->src[2] = MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), t_swizzle(GET_SWZ(src[0].Swizzle, 1)),	// y
-					 t_swizzle(GET_SWZ(src[0].Swizzle, 0)),	// x
-					 VSF_IN_COMPONENT_ZERO,	// z
-					 t_swizzle(GET_SWZ(src[0].Swizzle, 3)),	// w
-					 t_src_class(src[0].File),
-					 src[0].
-					 NegateBase ? VSF_FLAG_ALL :
-					 VSF_FLAG_NONE) | (src[0].
-							   RelAddr << 4);
+	inst[1] = MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), t_swizzle(GET_SWZ(src[0].Swizzle, 0)),	// x
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 3)),	// w
+				  VSF_IN_COMPONENT_ZERO,	// z
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 1)),	// y
+				  t_src_class(src[0].File),
+				  src[0].
+				  NegateBase ? VSF_FLAG_ALL :
+				  VSF_FLAG_NONE) | (src[0].RelAddr << 4);
+	inst[2] = MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), t_swizzle(GET_SWZ(src[0].Swizzle, 1)),	// y
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 3)),	// w
+				  VSF_IN_COMPONENT_ZERO,	// z
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 0)),	// x
+				  t_src_class(src[0].File),
+				  src[0].
+				  NegateBase ? VSF_FLAG_ALL :
+				  VSF_FLAG_NONE) | (src[0].RelAddr << 4);
+	inst[3] = MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), t_swizzle(GET_SWZ(src[0].Swizzle, 1)),	// y
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 0)),	// x
+				  VSF_IN_COMPONENT_ZERO,	// z
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 3)),	// w
+				  t_src_class(src[0].File),
+				  src[0].
+				  NegateBase ? VSF_FLAG_ALL :
+				  VSF_FLAG_NONE) | (src[0].RelAddr << 4);
+
+	return inst;
 }
 
-static void t_opcode_log(struct r300_vertex_program *vp,
-			 struct prog_instruction *vpi,
-			 struct r300_vertprog_instruction *o_inst,
-			 struct prog_src_register src[3])
+static GLuint *t_opcode_log(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
 {
-	o_inst->opcode =
+	inst[0] =
 	    MAKE_VSF_OP(R300_VPI_OUT_OP_LOG, t_dst_index(vp, &vpi->DstReg),
 			t_dst_mask(vpi->DstReg.WriteMask),
 			t_dst_class(vpi->DstReg.File));
 
-	o_inst->src[0] = t_src_scalar(vp, &src[0]);
-	o_inst->src[1] = ZERO_SRC_0;
-	o_inst->src[2] = ZERO_SRC_0;
+	inst[1] = t_src_scalar(vp, &src[0]);
+	inst[2] = ZERO_SRC_0;
+	inst[3] = ZERO_SRC_0;
+
+	return inst;
 }
 
-static void t_opcode_mad(struct r300_vertex_program *vp,
-			 struct prog_instruction *vpi,
-			 struct r300_vertprog_instruction *o_inst,
-			 struct prog_src_register src[3])
+static GLuint *t_opcode_mad(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
 {
 	unsigned long hw_op;
 
@@ -708,80 +717,84 @@ static void t_opcode_mad(struct r300_vertex_program *vp,
 		 PROGRAM_TEMPORARY) ? R300_VPI_OUT_OP_MAD_2 :
 	    R300_VPI_OUT_OP_MAD;
 
-	o_inst->opcode =
+	inst[0] =
 	    MAKE_VSF_OP(hw_op, t_dst_index(vp, &vpi->DstReg),
 			t_dst_mask(vpi->DstReg.WriteMask),
 			t_dst_class(vpi->DstReg.File));
-	o_inst->src[0] = t_src(vp, &src[0]);
-	o_inst->src[1] = t_src(vp, &src[1]);
-	o_inst->src[2] = t_src(vp, &src[2]);
+	inst[1] = t_src(vp, &src[0]);
+	inst[2] = t_src(vp, &src[1]);
+	inst[3] = t_src(vp, &src[2]);
+
+	return inst;
 }
 
-static void t_opcode_max(struct r300_vertex_program *vp,
-			 struct prog_instruction *vpi,
-			 struct r300_vertprog_instruction *o_inst,
-			 struct prog_src_register src[3])
+static GLuint *t_opcode_max(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
 {
-	o_inst->opcode =
+	inst[0] =
 	    MAKE_VSF_OP(R300_VPI_OUT_OP_MAX, t_dst_index(vp, &vpi->DstReg),
 			t_dst_mask(vpi->DstReg.WriteMask),
 			t_dst_class(vpi->DstReg.File));
 
-	o_inst->src[0] = t_src(vp, &src[0]);
-	o_inst->src[1] = t_src(vp, &src[1]);
-	o_inst->src[2] = ZERO_SRC_1;
+	inst[1] = t_src(vp, &src[0]);
+	inst[2] = t_src(vp, &src[1]);
+	inst[3] = ZERO_SRC_1;
+
+	return inst;
 }
 
-static void t_opcode_min(struct r300_vertex_program *vp,
-			 struct prog_instruction *vpi,
-			 struct r300_vertprog_instruction *o_inst,
-			 struct prog_src_register src[3])
+static GLuint *t_opcode_min(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
 {
-	o_inst->opcode =
+	inst[0] =
 	    MAKE_VSF_OP(R300_VPI_OUT_OP_MIN, t_dst_index(vp, &vpi->DstReg),
 			t_dst_mask(vpi->DstReg.WriteMask),
 			t_dst_class(vpi->DstReg.File));
 
-	o_inst->src[0] = t_src(vp, &src[0]);
-	o_inst->src[1] = t_src(vp, &src[1]);
-	o_inst->src[2] = ZERO_SRC_1;
+	inst[1] = t_src(vp, &src[0]);
+	inst[2] = t_src(vp, &src[1]);
+	inst[3] = ZERO_SRC_1;
+
+	return inst;
 }
 
-static void t_opcode_mov(struct r300_vertex_program *vp,
-			 struct prog_instruction *vpi,
-			 struct r300_vertprog_instruction *o_inst,
-			 struct prog_src_register src[3])
+static GLuint *t_opcode_mov(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
 {
 	//ADD RESULT 1.X Y Z W PARAM 0{} {X Y Z W} PARAM 0{} {ZERO ZERO ZERO ZERO}
 
 #if 1
-	o_inst->opcode =
+	inst[0] =
 	    MAKE_VSF_OP(R300_VPI_OUT_OP_ADD, t_dst_index(vp, &vpi->DstReg),
 			t_dst_mask(vpi->DstReg.WriteMask),
 			t_dst_class(vpi->DstReg.File));
-	o_inst->src[0] = t_src(vp, &src[0]);
-	o_inst->src[1] = ZERO_SRC_0;
-	o_inst->src[2] = ZERO_SRC_0;
+	inst[1] = t_src(vp, &src[0]);
+	inst[2] = ZERO_SRC_0;
+	inst[3] = ZERO_SRC_0;
 #else
 	hw_op =
 	    (src[0].File ==
 	     PROGRAM_TEMPORARY) ? R300_VPI_OUT_OP_MAD_2 :
 	    R300_VPI_OUT_OP_MAD;
 
-	o_inst->opcode =
+	inst[0] =
 	    MAKE_VSF_OP(hw_op, t_dst_index(vp, &vpi->DstReg),
 			t_dst_mask(vpi->DstReg.WriteMask),
 			t_dst_class(vpi->DstReg.File));
-	o_inst->src[0] = t_src(vp, &src[0]);
-	o_inst->src[1] = ONE_SRC_0;
-	o_inst->src[2] = ZERO_SRC_0;
+	inst[1] = t_src(vp, &src[0]);
+	inst[2] = ONE_SRC_0;
+	inst[3] = ZERO_SRC_0;
 #endif
+
+	return inst;
 }
 
-static void t_opcode_mul(struct r300_vertex_program *vp,
-			 struct prog_instruction *vpi,
-			 struct r300_vertprog_instruction *o_inst,
-			 struct prog_src_register src[3])
+static GLuint *t_opcode_mul(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
 {
 	unsigned long hw_op;
 
@@ -792,94 +805,100 @@ static void t_opcode_mul(struct r300_vertex_program *vp,
 		 PROGRAM_TEMPORARY) ? R300_VPI_OUT_OP_MAD_2 :
 	    R300_VPI_OUT_OP_MAD;
 
-	o_inst->opcode =
+	inst[0] =
 	    MAKE_VSF_OP(hw_op, t_dst_index(vp, &vpi->DstReg),
 			t_dst_mask(vpi->DstReg.WriteMask),
 			t_dst_class(vpi->DstReg.File));
-	o_inst->src[0] = t_src(vp, &src[0]);
-	o_inst->src[1] = t_src(vp, &src[1]);
+	inst[1] = t_src(vp, &src[0]);
+	inst[2] = t_src(vp, &src[1]);
+
+	inst[3] = ZERO_SRC_1;
 
-	o_inst->src[2] = ZERO_SRC_1;
+	return inst;
 }
 
-static void t_opcode_pow(struct r300_vertex_program *vp,
-			 struct prog_instruction *vpi,
-			 struct r300_vertprog_instruction *o_inst,
-			 struct prog_src_register src[3])
+static GLuint *t_opcode_pow(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
 {
-	o_inst->opcode =
+	inst[0] =
 	    MAKE_VSF_OP(R300_VPI_OUT_OP_POW, t_dst_index(vp, &vpi->DstReg),
 			t_dst_mask(vpi->DstReg.WriteMask),
 			t_dst_class(vpi->DstReg.File));
-	o_inst->src[0] = t_src_scalar(vp, &src[0]);
-	o_inst->src[1] = ZERO_SRC_0;
-	o_inst->src[2] = t_src_scalar(vp, &src[1]);
+	inst[1] = t_src_scalar(vp, &src[0]);
+	inst[2] = ZERO_SRC_0;
+	inst[3] = t_src_scalar(vp, &src[1]);
+
+	return inst;
 }
 
-static void t_opcode_rcp(struct r300_vertex_program *vp,
-			 struct prog_instruction *vpi,
-			 struct r300_vertprog_instruction *o_inst,
-			 struct prog_src_register src[3])
+static GLuint *t_opcode_rcp(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
 {
-	o_inst->opcode =
+	inst[0] =
 	    MAKE_VSF_OP(R300_VPI_OUT_OP_RCP, t_dst_index(vp, &vpi->DstReg),
 			t_dst_mask(vpi->DstReg.WriteMask),
 			t_dst_class(vpi->DstReg.File));
 
-	o_inst->src[0] = t_src_scalar(vp, &src[0]);
-	o_inst->src[1] = ZERO_SRC_0;
-	o_inst->src[2] = ZERO_SRC_0;
+	inst[1] = t_src_scalar(vp, &src[0]);
+	inst[2] = ZERO_SRC_0;
+	inst[3] = ZERO_SRC_0;
+
+	return inst;
 }
 
-static void t_opcode_rsq(struct r300_vertex_program *vp,
-			 struct prog_instruction *vpi,
-			 struct r300_vertprog_instruction *o_inst,
-			 struct prog_src_register src[3])
+static GLuint *t_opcode_rsq(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
 {
-	o_inst->opcode =
+	inst[0] =
 	    MAKE_VSF_OP(R300_VPI_OUT_OP_RSQ, t_dst_index(vp, &vpi->DstReg),
 			t_dst_mask(vpi->DstReg.WriteMask),
 			t_dst_class(vpi->DstReg.File));
 
-	o_inst->src[0] = t_src_scalar(vp, &src[0]);
-	o_inst->src[1] = ZERO_SRC_0;
-	o_inst->src[2] = ZERO_SRC_0;
+	inst[1] = t_src_scalar(vp, &src[0]);
+	inst[2] = ZERO_SRC_0;
+	inst[3] = ZERO_SRC_0;
+
+	return inst;
 }
 
-static void t_opcode_sge(struct r300_vertex_program *vp,
-			 struct prog_instruction *vpi,
-			 struct r300_vertprog_instruction *o_inst,
-			 struct prog_src_register src[3])
+static GLuint *t_opcode_sge(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
 {
-	o_inst->opcode =
+	inst[0] =
 	    MAKE_VSF_OP(R300_VPI_OUT_OP_SGE, t_dst_index(vp, &vpi->DstReg),
 			t_dst_mask(vpi->DstReg.WriteMask),
 			t_dst_class(vpi->DstReg.File));
 
-	o_inst->src[0] = t_src(vp, &src[0]);
-	o_inst->src[1] = t_src(vp, &src[1]);
-	o_inst->src[2] = ZERO_SRC_1;
+	inst[1] = t_src(vp, &src[0]);
+	inst[2] = t_src(vp, &src[1]);
+	inst[3] = ZERO_SRC_1;
+
+	return inst;
 }
 
-static void t_opcode_slt(struct r300_vertex_program *vp,
-			 struct prog_instruction *vpi,
-			 struct r300_vertprog_instruction *o_inst,
-			 struct prog_src_register src[3])
+static GLuint *t_opcode_slt(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
 {
-	o_inst->opcode =
+	inst[0] =
 	    MAKE_VSF_OP(R300_VPI_OUT_OP_SLT, t_dst_index(vp, &vpi->DstReg),
 			t_dst_mask(vpi->DstReg.WriteMask),
 			t_dst_class(vpi->DstReg.File));
 
-	o_inst->src[0] = t_src(vp, &src[0]);
-	o_inst->src[1] = t_src(vp, &src[1]);
-	o_inst->src[2] = ZERO_SRC_1;
+	inst[1] = t_src(vp, &src[0]);
+	inst[2] = t_src(vp, &src[1]);
+	inst[3] = ZERO_SRC_1;
+
+	return inst;
 }
 
-static void t_opcode_sub(struct r300_vertex_program *vp,
-			 struct prog_instruction *vpi,
-			 struct r300_vertprog_instruction *o_inst,
-			 struct prog_src_register src[3])
+static GLuint *t_opcode_sub(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
 {
 	unsigned long hw_op;
 
@@ -891,13 +910,13 @@ static void t_opcode_sub(struct r300_vertex_program *vp,
 		 PROGRAM_TEMPORARY) ? R300_VPI_OUT_OP_MAD_2 :
 	    R300_VPI_OUT_OP_MAD;
 
-	o_inst->opcode =
+	inst[0] =
 	    MAKE_VSF_OP(hw_op, t_dst_index(vp, &vpi->DstReg),
 			t_dst_mask(vpi->DstReg.WriteMask),
 			t_dst_class(vpi->DstReg.File));
-	o_inst->src[0] = t_src(vp, &src[0]);
-	o_inst->src[1] = ONE_SRC_0;
-	o_inst->src[2] =
+	inst[1] = t_src(vp, &src[0]);
+	inst[2] = ONE_SRC_0;
+	inst[3] =
 	    MAKE_VSF_SOURCE(t_src_index(vp, &src[1]),
 			    t_swizzle(GET_SWZ(src[1].Swizzle, 0)),
 			    t_swizzle(GET_SWZ(src[1].Swizzle, 1)),
@@ -908,13 +927,13 @@ static void t_opcode_sub(struct r300_vertex_program *vp,
 			     NegateBase) ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
 	    (src[1].RelAddr << 4);
 #else
-	o_inst->opcode =
+	inst[0] =
 	    MAKE_VSF_OP(R300_VPI_OUT_OP_ADD, t_dst_index(vp, &vpi->DstReg),
 			t_dst_mask(vpi->DstReg.WriteMask),
 			t_dst_class(vpi->DstReg.File));
 
-	o_inst->src[0] = t_src(vp, &src[0]);
-	o_inst->src[1] =
+	inst[1] = t_src(vp, &src[0]);
+	inst[2] =
 	    MAKE_VSF_SOURCE(t_src_index(vp, &src[1]),
 			    t_swizzle(GET_SWZ(src[1].Swizzle, 0)),
 			    t_swizzle(GET_SWZ(src[1].Swizzle, 1)),
@@ -924,113 +943,110 @@ static void t_opcode_sub(struct r300_vertex_program *vp,
 			    (!src[1].
 			     NegateBase) ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
 	    (src[1].RelAddr << 4);
-	o_inst->src[2] = 0;
+	inst[3] = 0;
 #endif
+
+	return inst;
 }
 
-static void t_opcode_swz(struct r300_vertex_program *vp,
-			 struct prog_instruction *vpi,
-			 struct r300_vertprog_instruction *o_inst,
-			 struct prog_src_register src[3])
+static GLuint *t_opcode_swz(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
 {
 	//ADD RESULT 1.X Y Z W PARAM 0{} {X Y Z W} PARAM 0{} {ZERO ZERO ZERO ZERO}
 
 #if 1
-	o_inst->opcode =
+	inst[0] =
 	    MAKE_VSF_OP(R300_VPI_OUT_OP_ADD, t_dst_index(vp, &vpi->DstReg),
 			t_dst_mask(vpi->DstReg.WriteMask),
 			t_dst_class(vpi->DstReg.File));
-	o_inst->src[0] = t_src(vp, &src[0]);
-	o_inst->src[1] = ZERO_SRC_0;
-	o_inst->src[2] = ZERO_SRC_0;
+	inst[1] = t_src(vp, &src[0]);
+	inst[2] = ZERO_SRC_0;
+	inst[3] = ZERO_SRC_0;
 #else
 	hw_op =
 	    (src[0].File ==
 	     PROGRAM_TEMPORARY) ? R300_VPI_OUT_OP_MAD_2 :
 	    R300_VPI_OUT_OP_MAD;
 
-	o_inst->opcode =
+	inst[0] =
 	    MAKE_VSF_OP(hw_op, t_dst_index(vp, &vpi->DstReg),
 			t_dst_mask(vpi->DstReg.WriteMask),
 			t_dst_class(vpi->DstReg.File));
-	o_inst->src[0] = t_src(vp, &src[0]);
-	o_inst->src[1] = ONE_SRC_0;
-	o_inst->src[2] = ZERO_SRC_0;
+	inst[1] = t_src(vp, &src[0]);
+	inst[2] = ONE_SRC_0;
+	inst[3] = ZERO_SRC_0;
 #endif
+
+	return inst;
 }
 
-static struct r300_vertprog_instruction *t_opcode_xpd(struct r300_vertex_program *vp,
-						      struct prog_instruction *vpi,
-						      struct r300_vertprog_instruction *o_inst,
-						      struct prog_src_register src[3],
-						      int *u_temp_i)
+static GLuint *t_opcode_xpd(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3], int *u_temp_i)
 {
 	/* mul r0, r1.yzxw, r2.zxyw
 	   mad r0, -r2.yzxw, r1.zxyw, r0
 	   NOTE: might need MAD_2
 	 */
 
-	o_inst->opcode =
+	inst[0] =
 	    MAKE_VSF_OP(R300_VPI_OUT_OP_MAD, *u_temp_i,
 			t_dst_mask(vpi->DstReg.WriteMask),
 			VSF_OUT_CLASS_TMP);
 
-	o_inst->src[0] = MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), t_swizzle(GET_SWZ(src[0].Swizzle, 1)),	// y
-					 t_swizzle(GET_SWZ(src[0].Swizzle, 2)),	// z
-					 t_swizzle(GET_SWZ(src[0].Swizzle, 0)),	// x
-					 t_swizzle(GET_SWZ(src[0].Swizzle, 3)),	// w
-					 t_src_class(src[0].File),
-					 src[0].
-					 NegateBase ? VSF_FLAG_ALL :
-					 VSF_FLAG_NONE) | (src[0].
-							   RelAddr << 4);
-
-	o_inst->src[1] = MAKE_VSF_SOURCE(t_src_index(vp, &src[1]), t_swizzle(GET_SWZ(src[1].Swizzle, 2)),	// z
-					 t_swizzle(GET_SWZ(src[1].Swizzle, 0)),	// x
-					 t_swizzle(GET_SWZ(src[1].Swizzle, 1)),	// y
-					 t_swizzle(GET_SWZ(src[1].Swizzle, 3)),	// w
-					 t_src_class(src[1].File),
-					 src[1].
-					 NegateBase ? VSF_FLAG_ALL :
-					 VSF_FLAG_NONE) | (src[1].
-							   RelAddr << 4);
-
-	o_inst->src[2] = ZERO_SRC_1;
-	o_inst++;
+	inst[1] = MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), t_swizzle(GET_SWZ(src[0].Swizzle, 1)),	// y
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 2)),	// z
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 0)),	// x
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 3)),	// w
+				  t_src_class(src[0].File),
+				  src[0].
+				  NegateBase ? VSF_FLAG_ALL :
+				  VSF_FLAG_NONE) | (src[0].RelAddr << 4);
+
+	inst[2] = MAKE_VSF_SOURCE(t_src_index(vp, &src[1]), t_swizzle(GET_SWZ(src[1].Swizzle, 2)),	// z
+				  t_swizzle(GET_SWZ(src[1].Swizzle, 0)),	// x
+				  t_swizzle(GET_SWZ(src[1].Swizzle, 1)),	// y
+				  t_swizzle(GET_SWZ(src[1].Swizzle, 3)),	// w
+				  t_src_class(src[1].File),
+				  src[1].
+				  NegateBase ? VSF_FLAG_ALL :
+				  VSF_FLAG_NONE) | (src[1].RelAddr << 4);
+
+	inst[3] = ZERO_SRC_1;
+	inst += 4;
 	(*u_temp_i)--;
 
-	o_inst->opcode =
+	inst[0] =
 	    MAKE_VSF_OP(R300_VPI_OUT_OP_MAD, t_dst_index(vp, &vpi->DstReg),
 			t_dst_mask(vpi->DstReg.WriteMask),
 			t_dst_class(vpi->DstReg.File));
 
-	o_inst->src[0] = MAKE_VSF_SOURCE(t_src_index(vp, &src[1]), t_swizzle(GET_SWZ(src[1].Swizzle, 1)),	// y
-					 t_swizzle(GET_SWZ(src[1].Swizzle, 2)),	// z
-					 t_swizzle(GET_SWZ(src[1].Swizzle, 0)),	// x
-					 t_swizzle(GET_SWZ(src[1].Swizzle, 3)),	// w
-					 t_src_class(src[1].File),
-					 (!src[1].
-					  NegateBase) ? VSF_FLAG_ALL :
-					 VSF_FLAG_NONE) | (src[1].
-							   RelAddr << 4);
-
-	o_inst->src[1] = MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), t_swizzle(GET_SWZ(src[0].Swizzle, 2)),	// z
-					 t_swizzle(GET_SWZ(src[0].Swizzle, 0)),	// x
-					 t_swizzle(GET_SWZ(src[0].Swizzle, 1)),	// y
-					 t_swizzle(GET_SWZ(src[0].Swizzle, 3)),	// w
-					 t_src_class(src[0].File),
-					 src[0].
-					 NegateBase ? VSF_FLAG_ALL :
-					 VSF_FLAG_NONE) | (src[0].
-							   RelAddr << 4);
-
-	o_inst->src[2] =
+	inst[1] = MAKE_VSF_SOURCE(t_src_index(vp, &src[1]), t_swizzle(GET_SWZ(src[1].Swizzle, 1)),	// y
+				  t_swizzle(GET_SWZ(src[1].Swizzle, 2)),	// z
+				  t_swizzle(GET_SWZ(src[1].Swizzle, 0)),	// x
+				  t_swizzle(GET_SWZ(src[1].Swizzle, 3)),	// w
+				  t_src_class(src[1].File),
+				  (!src[1].
+				   NegateBase) ? VSF_FLAG_ALL :
+				  VSF_FLAG_NONE) | (src[1].RelAddr << 4);
+
+	inst[2] = MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), t_swizzle(GET_SWZ(src[0].Swizzle, 2)),	// z
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 0)),	// x
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 1)),	// y
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 3)),	// w
+				  t_src_class(src[0].File),
+				  src[0].
+				  NegateBase ? VSF_FLAG_ALL :
+				  VSF_FLAG_NONE) | (src[0].RelAddr << 4);
+
+	inst[3] =
 	    MAKE_VSF_SOURCE(*u_temp_i + 1, VSF_IN_COMPONENT_X,
 			    VSF_IN_COMPONENT_Y, VSF_IN_COMPONENT_Z,
 			    VSF_IN_COMPONENT_W, VSF_IN_CLASS_TMP,
 			    VSF_FLAG_NONE);
 
-	return o_inst;
+	return inst;
 }
 
 static void t_inputs_outputs(struct r300_vertex_program *vp)
@@ -1067,7 +1083,7 @@ static void t_inputs_outputs(struct r300_vertex_program *vp)
 	if (vp->key.OutputsWritten & (1 << VERT_RESULT_BFC0)) {
 		vp->outputs[VERT_RESULT_BFC0] =
 		    vp->outputs[VERT_RESULT_COL0] + 2;
-		cur_reg = vp->outputs[VERT_RESULT_BFC0] + 1;
+		cur_reg = vp->outputs[VERT_RESULT_BFC0] + 2;
 	}
 
 	if (vp->key.OutputsWritten & (1 << VERT_RESULT_BFC1)) {
@@ -1092,7 +1108,7 @@ static void r300TranslateVertexShader(struct r300_vertex_program *vp,
 				      struct prog_instruction *vpi)
 {
 	int i;
-	struct r300_vertprog_instruction *o_inst;
+	GLuint *inst;
 	unsigned long num_operands;
 	/* Initial value should be last tmp reg that hw supports.
 	   Strangely enough r300 doesnt mind even though these would be out of range.
@@ -1108,8 +1124,8 @@ static void r300TranslateVertexShader(struct r300_vertex_program *vp,
 
 	t_inputs_outputs(vp);
 
-	for (o_inst = vp->program.body.i; vpi->Opcode != OPCODE_END;
-	     vpi++, o_inst++) {
+	for (inst = vp->program.body.i; vpi->Opcode != OPCODE_END;
+	     vpi++, inst += 4) {
 
 		FREE_TEMPS();
 
@@ -1129,12 +1145,12 @@ static void r300TranslateVertexShader(struct r300_vertex_program *vp,
 		if (num_operands == 3) {	/* TODO: scalars */
 			if (CMP_SRCS(src[1], src[2])
 			    || CMP_SRCS(src[0], src[2])) {
-				o_inst->opcode =
+				inst[0] =
 				    MAKE_VSF_OP(R300_VPI_OUT_OP_ADD,
 						u_temp_i, VSF_FLAG_ALL,
 						VSF_OUT_CLASS_TMP);
 
-				o_inst->src[0] =
+				inst[1] =
 				    MAKE_VSF_SOURCE(t_src_index
 						    (vp, &src[2]),
 						    SWIZZLE_X, SWIZZLE_Y,
@@ -1144,9 +1160,9 @@ static void r300TranslateVertexShader(struct r300_vertex_program *vp,
 						    VSF_FLAG_NONE) |
 				    (src[2].RelAddr << 4);
 
-				o_inst->src[1] = ZERO_SRC_2;
-				o_inst->src[2] = ZERO_SRC_2;
-				o_inst++;
+				inst[2] = ZERO_SRC_2;
+				inst[3] = ZERO_SRC_2;
+				inst += 4;
 
 				src[2].File = PROGRAM_TEMPORARY;
 				src[2].Index = u_temp_i;
@@ -1157,12 +1173,12 @@ static void r300TranslateVertexShader(struct r300_vertex_program *vp,
 
 		if (num_operands >= 2) {
 			if (CMP_SRCS(src[1], src[0])) {
-				o_inst->opcode =
+				inst[0] =
 				    MAKE_VSF_OP(R300_VPI_OUT_OP_ADD,
 						u_temp_i, VSF_FLAG_ALL,
 						VSF_OUT_CLASS_TMP);
 
-				o_inst->src[0] =
+				inst[1] =
 				    MAKE_VSF_SOURCE(t_src_index
 						    (vp, &src[0]),
 						    SWIZZLE_X, SWIZZLE_Y,
@@ -1172,9 +1188,9 @@ static void r300TranslateVertexShader(struct r300_vertex_program *vp,
 						    VSF_FLAG_NONE) |
 				    (src[0].RelAddr << 4);
 
-				o_inst->src[1] = ZERO_SRC_0;
-				o_inst->src[2] = ZERO_SRC_0;
-				o_inst++;
+				inst[2] = ZERO_SRC_0;
+				inst[3] = ZERO_SRC_0;
+				inst += 4;
 
 				src[0].File = PROGRAM_TEMPORARY;
 				src[0].Index = u_temp_i;
@@ -1185,87 +1201,89 @@ static void r300TranslateVertexShader(struct r300_vertex_program *vp,
 
 		switch (vpi->Opcode) {
 		case OPCODE_ABS:
-			t_opcode_abs(vp, vpi, o_inst, src);
+			inst = t_opcode_abs(vp, vpi, inst, src);
 			break;
 		case OPCODE_ADD:
-			t_opcode_add(vp, vpi, o_inst, src);
+			inst = t_opcode_add(vp, vpi, inst, src);
 			break;
 		case OPCODE_ARL:
-			t_opcode_arl(vp, vpi, o_inst, src);
+			inst = t_opcode_arl(vp, vpi, inst, src);
 			break;
 		case OPCODE_DP3:
-			t_opcode_dp3(vp, vpi, o_inst, src);
+			inst = t_opcode_dp3(vp, vpi, inst, src);
 			break;
 		case OPCODE_DP4:
-			t_opcode_dp4(vp, vpi, o_inst, src);
+			inst = t_opcode_dp4(vp, vpi, inst, src);
 			break;
 		case OPCODE_DPH:
-			t_opcode_dph(vp, vpi, o_inst, src);
+			inst = t_opcode_dph(vp, vpi, inst, src);
 			break;
 		case OPCODE_DST:
-			t_opcode_dst(vp, vpi, o_inst, src);
+			inst = t_opcode_dst(vp, vpi, inst, src);
 			break;
 		case OPCODE_EX2:
-			t_opcode_ex2(vp, vpi, o_inst, src);
+			inst = t_opcode_ex2(vp, vpi, inst, src);
 			break;
 		case OPCODE_EXP:
-			t_opcode_exp(vp, vpi, o_inst, src);
+			inst = t_opcode_exp(vp, vpi, inst, src);
 			break;
 		case OPCODE_FLR:
-			/* FIXME */
-			o_inst = t_opcode_flr(vp, vpi, o_inst, src, &u_temp_i);
+			inst =
+			    t_opcode_flr(vp, vpi, inst, src, /* FIXME */
+					 &u_temp_i);
 			break;
 		case OPCODE_FRC:
-			t_opcode_frc(vp, vpi, o_inst, src);
+			inst = t_opcode_frc(vp, vpi, inst, src);
 			break;
 		case OPCODE_LG2:
-			t_opcode_lg2(vp, vpi, o_inst, src);
+			inst = t_opcode_lg2(vp, vpi, inst, src);
 			break;
 		case OPCODE_LIT:
-			t_opcode_lit(vp, vpi, o_inst, src);
+			inst = t_opcode_lit(vp, vpi, inst, src);
 			break;
 		case OPCODE_LOG:
-			t_opcode_log(vp, vpi, o_inst, src);
+			inst = t_opcode_log(vp, vpi, inst, src);
 			break;
 		case OPCODE_MAD:
-			t_opcode_mad(vp, vpi, o_inst, src);
+			inst = t_opcode_mad(vp, vpi, inst, src);
 			break;
 		case OPCODE_MAX:
-			t_opcode_max(vp, vpi, o_inst, src);
+			inst = t_opcode_max(vp, vpi, inst, src);
 			break;
 		case OPCODE_MIN:
-			t_opcode_min(vp, vpi, o_inst, src);
+			inst = t_opcode_min(vp, vpi, inst, src);
 			break;
 		case OPCODE_MOV:
-			t_opcode_mov(vp, vpi, o_inst, src);
+			inst = t_opcode_mov(vp, vpi, inst, src);
 			break;
 		case OPCODE_MUL:
-			t_opcode_mul(vp, vpi, o_inst, src);
+			inst = t_opcode_mul(vp, vpi, inst, src);
 			break;
 		case OPCODE_POW:
-			t_opcode_pow(vp, vpi, o_inst, src);
+			inst = t_opcode_pow(vp, vpi, inst, src);
 			break;
 		case OPCODE_RCP:
-			t_opcode_rcp(vp, vpi, o_inst, src);
+			inst = t_opcode_rcp(vp, vpi, inst, src);
 			break;
 		case OPCODE_RSQ:
-			t_opcode_rsq(vp, vpi, o_inst, src);
+			inst = t_opcode_rsq(vp, vpi, inst, src);
 			break;
 		case OPCODE_SGE:
-			t_opcode_sge(vp, vpi, o_inst, src);
+			inst = t_opcode_sge(vp, vpi, inst, src);
 			break;
 		case OPCODE_SLT:
-			t_opcode_slt(vp, vpi, o_inst, src);
+			inst = t_opcode_slt(vp, vpi, inst, src);
 			break;
 		case OPCODE_SUB:
-			t_opcode_sub(vp, vpi, o_inst, src);
+			inst = t_opcode_sub(vp, vpi, inst, src);
 			break;
 		case OPCODE_SWZ:
-			t_opcode_swz(vp, vpi, o_inst, src);
+			inst = t_opcode_swz(vp, vpi, inst, src);
 			break;
 		case OPCODE_XPD:
-			/* FIXME */
-			o_inst = t_opcode_xpd(vp, vpi, o_inst, src, &u_temp_i);
+			inst =
+			    t_opcode_xpd(vp, vpi, inst, src, /* FIXME */
+					 &u_temp_i);
 			break;
 		default:
 			assert(0);
@@ -1273,7 +1291,7 @@ static void r300TranslateVertexShader(struct r300_vertex_program *vp,
 		}
 	}
 
-	vp->program.length = (o_inst - vp->program.body.i) * 4;
+	vp->program.length = (inst - vp->program.body.i);
 	if (vp->program.length >= VSF_MAX_FRAGMENT_LENGTH) {
 		vp->program.length = 0;
 		vp->native = GL_FALSE;
diff --git a/src/mesa/drivers/dri/r300/r300_vertprog.h b/src/mesa/drivers/dri/r300/r300_vertprog.h
index 2d399e243a..3df0eee799 100644
--- a/src/mesa/drivers/dri/r300/r300_vertprog.h
+++ b/src/mesa/drivers/dri/r300/r300_vertprog.h
@@ -3,11 +3,6 @@
 
 #include "r300_reg.h"
 
-struct r300_vertprog_instruction {
-	GLuint opcode;
-	GLuint src[3];
-};
-
 #define VSF_FLAG_X	1
 #define VSF_FLAG_Y	2
 #define VSF_FLAG_Z	4
diff --git a/src/mesa/drivers/dri/savage/savagetris.c b/src/mesa/drivers/dri/savage/savagetris.c
index 3dd821a4d3..4ce2f60b4f 100644
--- a/src/mesa/drivers/dri/savage/savagetris.c
+++ b/src/mesa/drivers/dri/savage/savagetris.c
@@ -131,7 +131,9 @@ static __inline__ void savage_draw_point (savageContextPtr imesa,
    u_int32_t *vb = savageAllocVtxBuf (imesa, 6*vertsize);
    const GLfloat x = tmp->v.x;
    const GLfloat y = tmp->v.y;
-   const GLfloat sz = imesa->glCtx->Point._Size * .5;
+   const GLfloat sz = 0.5 * CLAMP(imesa->glCtx->Point.Size,
+                                  imesa->glCtx->Const.MinPointSize,
+                                  imesa->glCtx->Const.MaxPointSize);
    GLuint j;
 
    *(float *)&vb[0] = x - sz;
@@ -164,7 +166,9 @@ static __inline__ void savage_draw_line (savageContextPtr imesa,
 					 savageVertexPtr v1 ) {
    GLuint vertsize = imesa->HwVertexSize;
    u_int32_t *vb = savageAllocVtxBuf (imesa, 6*vertsize);
-   GLfloat width = imesa->glCtx->Line._Width;
+   const GLfloat width = CLAMP(imesa->glCtx->Line.Width,
+                               imesa->glCtx->Const.MinLineWidth,
+                               imesa->glCtx->Const.MaxLineWidth);
    GLfloat dx, dy, ix, iy;
    GLuint j;
 
@@ -234,7 +238,9 @@ static __inline__ void savage_ptex_line (savageContextPtr imesa,
 					 savageVertexPtr v1 ) {
    GLuint vertsize = imesa->HwVertexSize;
    u_int32_t *vb = savageAllocVtxBuf (imesa, 6*vertsize);
-   GLfloat width = imesa->glCtx->Line._Width;
+   const GLfloat width = CLAMP(imesa->glCtx->Line.Width,
+                               imesa->glCtx->Const.MinLineWidth,
+                               imesa->glCtx->Const.MaxLineWidth);
    GLfloat dx, dy, ix, iy;
    savageVertex tmp0, tmp1;
    GLuint j;
@@ -281,7 +287,9 @@ static __inline__ void savage_ptex_point (savageContextPtr imesa,
    u_int32_t *vb = savageAllocVtxBuf (imesa, 6*vertsize);
    const GLfloat x = v0->v.x;
    const GLfloat y = v0->v.y;
-   const GLfloat sz = imesa->glCtx->Point._Size * .5;
+   const GLfloat sz = 0.5 * CLAMP(imesa->glCtx->Point.Size,
+                                  imesa->glCtx->Const.MinPointSize,
+                                  imesa->glCtx->Const.MaxPointSize);
    savageVertex tmp;
    GLuint j;
 
diff --git a/src/mesa/drivers/dri/tdfx/tdfx_tris.c b/src/mesa/drivers/dri/tdfx/tdfx_tris.c
index 96f9ae27fc..7252a7e7dc 100644
--- a/src/mesa/drivers/dri/tdfx/tdfx_tris.c
+++ b/src/mesa/drivers/dri/tdfx/tdfx_tris.c
@@ -184,7 +184,7 @@ tdfx_translate_vertex( GLcontext *ctx, const tdfxVertex *src, SWvertex *dst)
       }
    }
 
-   dst->pointSize = ctx->Point._Size;
+   dst->pointSize = ctx->Point.Size;
 }
 
 
diff --git a/src/mesa/drivers/windows/gdi/wgl.c b/src/mesa/drivers/windows/gdi/wgl.c
index fb23d210db..dad3dc1160 100644
--- a/src/mesa/drivers/windows/gdi/wgl.c
+++ b/src/mesa/drivers/windows/gdi/wgl.c
@@ -33,15 +33,30 @@
 /* We're essentially building part of GDI here, so define this so that
  * we get the right export linkage. */
 #ifdef __MINGW32__
-#include <GL/gl.h>
+
+#include <stdarg.h>
+#include <windef.h>
+#include <wincon.h>
+#include <winbase.h>
+
+#  if defined(BUILD_GL32)
+#    define WINGDIAPI __declspec(dllexport)	
+#  else
+#    define __W32API_USE_DLLIMPORT__
+#  endif
+
+#include <wingdi.h>
+#include "GL/mesa_wgl.h"
 #include <stdlib.h>
+
 #else
+
 #define _GDI32_
-#endif
 #include <windows.h>
 
-#include "glapi.h"
+#endif
 
+#include "glapi.h"
 #include "GL/wmesa.h"   /* protos for wmesa* functions */
 
 /*
@@ -339,7 +354,7 @@ WINGDIAPI int GLAPIENTRY wglGetPixelFormat(HDC hdc)
 }
 
 WINGDIAPI BOOL GLAPIENTRY wglSetPixelFormat(HDC hdc,int iPixelFormat,
-					    const PIXELFORMATDESCRIPTOR *ppfd)
+					const PIXELFORMATDESCRIPTOR *ppfd)
 {
     (void) hdc;
     
@@ -392,12 +407,12 @@ static BOOL wglUseFontBitmaps_FX(HDC fontDevice, DWORD firstChar,
     
     bitDevice = CreateCompatibleDC(fontDevice);
     
-    // Swap fore and back colors so the bitmap has the right polarity
+    /* Swap fore and back colors so the bitmap has the right polarity */
     tempColor = GetBkColor(bitDevice);
     SetBkColor(bitDevice, GetTextColor(bitDevice));
     SetTextColor(bitDevice, tempColor);
     
-    // Place chars based on base line
+    /* Place chars based on base line */
     VERIFY(SetTextAlign(bitDevice, TA_BASELINE) != GDI_ERROR ? 1 : 0);
     
     for(i = 0; i < (int)numChars; i++) {
@@ -410,36 +425,36 @@ static BOOL wglUseFontBitmaps_FX(HDC fontDevice, DWORD firstChar,
 	
 	curChar = (char)(i + firstChar);
 	
-	// Find how high/wide this character is
+	/* Find how high/wide this character is */
 	VERIFY(GetTextExtentPoint32(bitDevice, &curChar, 1, &size));
 	
-	// Create the output bitmap
+	/* Create the output bitmap */
 	charWidth = size.cx;
 	charHeight = size.cy;
-	// Round up to the next multiple of 32 bits
+	/* Round up to the next multiple of 32 bits */
 	bmapWidth = ((charWidth + 31) / 32) * 32;   
 	bmapHeight = charHeight;
 	bitObject = CreateCompatibleBitmap(bitDevice,
 					   bmapWidth,
 					   bmapHeight);
-	//VERIFY(bitObject);
+	/* VERIFY(bitObject); */
 	
-	// Assign the output bitmap to the device
+	/* Assign the output bitmap to the device */
 	origBmap = SelectObject(bitDevice, bitObject);
 	(void) VERIFY(origBmap);
 	
 	VERIFY( PatBlt( bitDevice, 0, 0, bmapWidth, bmapHeight,BLACKNESS ) );
 	
-	// Use our source font on the device
+	/* Use our source font on the device */
 	VERIFY(SelectObject(bitDevice, GetCurrentObject(fontDevice,OBJ_FONT)));
 	
-	// Draw the character
+	/* Draw the character */
 	VERIFY(TextOut(bitDevice, 0, metric.tmAscent, &curChar, 1));
 	
-	// Unselect our bmap object
+	/* Unselect our bmap object */
 	VERIFY(SelectObject(bitDevice, origBmap));
 	
-	// Convert the display dependant representation to a 1 bit deep DIB
+	/* Convert the display dependant representation to a 1 bit deep DIB */
 	numBytes = (bmapWidth * bmapHeight) / 8;
 	bmap = malloc(numBytes);
 	dibInfo->bmiHeader.biWidth = bmapWidth;
@@ -447,24 +462,24 @@ static BOOL wglUseFontBitmaps_FX(HDC fontDevice, DWORD firstChar,
 	res = GetDIBits(bitDevice, bitObject, 0, bmapHeight, bmap,
 			dibInfo,
 			DIB_RGB_COLORS);
-	//VERIFY(res);
+	/* VERIFY(res); */
 	
-	// Create the GL object
+	/* Create the GL object */
 	glNewList(i + listBase, GL_COMPILE);
 	glBitmap(bmapWidth, bmapHeight, 0.0, (GLfloat)metric.tmDescent,
 		 (GLfloat)charWidth, 0.0,
 		 bmap);
 	glEndList();
-	// CheckGL();
+	/* CheckGL(); */
 	
-	// Destroy the bmap object
+	/* Destroy the bmap object */
 	DeleteObject(bitObject);
 	
-	// Deallocate the bitmap data
+	/* Deallocate the bitmap data */
 	free(bmap);
     }
     
-    // Destroy the DC
+    /* Destroy the DC */
     VERIFY(DeleteDC(bitDevice));
     
     free(dibInfo);
diff --git a/src/mesa/drivers/windows/gdi/wmesa.c b/src/mesa/drivers/windows/gdi/wmesa.c
index 2eec188912..5b67439f0f 100644
--- a/src/mesa/drivers/windows/gdi/wmesa.c
+++ b/src/mesa/drivers/windows/gdi/wmesa.c
@@ -6,6 +6,7 @@
 #include "wmesadef.h"
 #include "colors.h"
 #include <GL/wmesa.h>
+#include <winuser.h>
 #include "context.h"
 #include "extensions.h"
 #include "framebuffer.h"
@@ -114,7 +115,7 @@ static void wmSetPixelFormat(WMesaFramebuffer pwfb, HDC hDC)
 {
     pwfb->cColorBits = GetDeviceCaps(hDC, BITSPIXEL);
 
-    // Only 16 and 32 bit targets are supported now
+    /* Only 16 and 32 bit targets are supported now */
     assert(pwfb->cColorBits == 0 ||
 	   pwfb->cColorBits == 16 || 
 	   pwfb->cColorBits == 32);
@@ -1171,7 +1172,7 @@ WMesaContext WMesaCreateContext(HDC hDC,
     /* I do not understand this contributed code */
     /* Support memory and device contexts */
     if(WindowFromDC(hDC) != NULL) {
-	c->hDC = GetDC(WindowFromDC(hDC)); // huh ????
+	c->hDC = GetDC(WindowFromDC(hDC)); /* huh ???? */
     }
     else {
 	c->hDC = hDC;
@@ -1404,6 +1405,7 @@ void WMesaSwapBuffers( HDC hdc )
  * table entries.  Hopefully, I'll find a better solution.  The
  * dispatch table generation scripts ought to be making these dummy
  * stubs as well. */
+#if !defined(__MINGW32__) || !defined(GL_NO_STDCALL)
 void gl_dispatch_stub_543(void){}
 void gl_dispatch_stub_544(void){}
 void gl_dispatch_stub_545(void){}
@@ -1471,3 +1473,4 @@ void gl_dispatch_stub_769(void){}
 void gl_dispatch_stub_770(void){}
 void gl_dispatch_stub_771(void){}
 
+#endif
diff --git a/src/mesa/drivers/windows/gdi/wmesadef.h b/src/mesa/drivers/windows/gdi/wmesadef.h
index 97b063a8ba..83a42e6082 100644
--- a/src/mesa/drivers/windows/gdi/wmesadef.h
+++ b/src/mesa/drivers/windows/gdi/wmesadef.h
@@ -1,6 +1,8 @@
 #ifndef WMESADEF_H
 #define WMESADEF_H
-
+#ifdef __MINGW32__
+#include <windows.h>
+#endif
 #include "context.h"
 
 
diff --git a/src/mesa/drivers/x11/fakeglx.c b/src/mesa/drivers/x11/fakeglx.c
index 1587df66bc..7a170b4d3d 100644
--- a/src/mesa/drivers/x11/fakeglx.c
+++ b/src/mesa/drivers/x11/fakeglx.c
@@ -296,6 +296,11 @@ save_glx_visual( Display *dpy, XVisualInfo *vinfo,
       }
    }
 
+   if (stereoFlag) {
+      /* stereo not supported */
+      return NULL;
+   }
+
    /* Comparing IDs uses less memory but sometimes fails. */
    /* XXX revisit this after 3.0 is finished. */
    if (_mesa_getenv("MESA_GLX_VISUAL_HACK"))
@@ -1079,7 +1084,7 @@ choose_visual( Display *dpy, int screen, const int *list, GLboolean fbConfig )
             else {
                stereo_flag = GL_TRUE;
             }
-            return NULL; /* stereo not supported */
+            break;
 	 case GLX_AUX_BUFFERS:
 	    parselist++;
             numAux = *parselist++;