6 files changed, 83 insertions, 60 deletions

diff --git a/src/mesa/drivers/dri/r200/r200_cmdbuf.c b/src/mesa/drivers/dri/r200/r200_cmdbuf.c
index f107ff5624..6d421f5d5e 100644
--- a/src/mesa/drivers/dri/r200/r200_cmdbuf.c
+++ b/src/mesa/drivers/dri/r200/r200_cmdbuf.c
@@ -47,17 +47,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "r200_sanity.h"
 #include "radeon_reg.h"
 
-static void print_state_atom( struct radeon_state_atom *state )
-{
-   int i;
-
-   fprintf(stderr, "emit %s/%d\n", state->name, state->cmd_size);
-
-   if (0 & R200_DEBUG & DEBUG_VERBOSE) 
-      for (i = 0 ; i < state->cmd_size ; i++) 
-	 fprintf(stderr, "\t%s[%d]: %x\n", state->name, i, state->cmd[i]);
-
-}
+#define DEBUG_CMDBUF         0
 
 /* The state atoms will be emitted in the order they appear in the atom list,
  * so this step is important.
@@ -141,7 +131,7 @@ static void r200SaveHwState( r200ContextPtr rmesa )
 	   rmesa->backup_store.cmd_used += size;
 	 }
 	 if (R200_DEBUG & DEBUG_STATE)
-	    print_state_atom( atom );
+	    radeon_print_state_atom( atom );
       }
    }
 
@@ -161,9 +151,9 @@ static INLINE void r200EmitAtoms(r200ContextPtr r200, GLboolean dirty)
      if ((atom->dirty || r200->hw.all_dirty) == dirty) {
        dwords = (*atom->check) (r200->radeon.glCtx, atom);
        if (dwords) {
-	 //	 if (DEBUG_CMDBUF && RADEON_DEBUG & DEBUG_STATE) {
-	 //	   r300PrintStateAtom(r300, atom);
-	 //		}
+	  if (DEBUG_CMDBUF && RADEON_DEBUG & DEBUG_STATE) {
+	     radeon_print_state_atom(atom);
+	  }
 	 if (atom->emit) {
 	   (*atom->emit)(r200->radeon.glCtx, atom);
 	 } else {
@@ -173,10 +163,10 @@ static INLINE void r200EmitAtoms(r200ContextPtr r200, GLboolean dirty)
 	 }
 	 atom->dirty = GL_FALSE;
        } else {
-	 //	 if (DEBUG_CMDBUF && RADEON_DEBUG & DEBUG_STATE) {
-	 //	   fprintf(stderr, "  skip state %s\n",
-	 //		   atom->name);
-	 //	 }
+	  if (DEBUG_CMDBUF && RADEON_DEBUG & DEBUG_STATE) {
+	     fprintf(stderr, "  skip state %s\n",
+		     atom->name);
+	  }
        }
      }
    }
diff --git a/src/mesa/drivers/dri/r200/r200_ioctl.c b/src/mesa/drivers/dri/r200/r200_ioctl.c
index 1bb97c5002..1fbbbfbb56 100644
--- a/src/mesa/drivers/dri/r200/r200_ioctl.c
+++ b/src/mesa/drivers/dri/r200/r200_ioctl.c
@@ -41,7 +41,9 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "main/context.h"
 #include "swrast/swrast.h"
 
+#include "radeon_cs.h"
 #include "r200_context.h"
+
 #include "common_cmdbuf.h"
 #include "r200_state.h"
 #include "r200_ioctl.h"
@@ -525,8 +527,8 @@ void r200Flush( GLcontext *ctx )
       rmesa->dma.flush( ctx );
 
    r200EmitState( rmesa );
-   
-   if (rmesa->store.cmd_used)
+
+   if (rmesa->radeon.cmdbuf.cs->cdw)
       rcommonFlushCmdBuf( &rmesa->radeon, __FUNCTION__ );
 }
 
diff --git a/src/mesa/drivers/dri/r200/r200_ioctl.h b/src/mesa/drivers/dri/r200/r200_ioctl.h
index f36a6fd90f..1ae0dd114b 100644
--- a/src/mesa/drivers/dri/r200/r200_ioctl.h
+++ b/src/mesa/drivers/dri/r200/r200_ioctl.h
@@ -201,40 +201,31 @@ static INLINE char *r200AllocCmdBuf( r200ContextPtr rmesa,
 }
 #endif
 
-static inline uint32_t cmdpacket3_clip(int cmd_type)
+static inline uint32_t cmdpacket3(int cmd_type)
 {
   drm_radeon_cmd_header_t cmd;
 
   cmd.i = 0;
-  cmd.header.cmd_type = RADEON_CMD_PACKET3_CLIP;
+  cmd.header.cmd_type = cmd_type;
 
   return (uint32_t)cmd.i;
 
 }
-#define OUT_BATCH_PACKET3_CLIP(packet, num_extra) do {	      \
+
+#define OUT_BATCH_PACKET3(packet, num_extra) do {	      \
     if (!b_l_rmesa->radeonScreen->kernel_mm) {		      \
-      OUT_BATCH(cmdpacket3_clip(0));			      \
-      OUT_BATCH(packet);				      \
+      OUT_BATCH(cmdpacket3(RADEON_CMD_PACKET3));				      \
+      OUT_BATCH(CP_PACKET3((packet), (num_extra)));	      \
     } else {						      \
       OUT_BATCH(CP_PACKET2);				      \
       OUT_BATCH(CP_PACKET3((packet), (num_extra)));	      \
     }							      \
   } while(0)
 
-static inline uint32_t cmdpacket3(int cmd_type)
-{
-  drm_radeon_cmd_header_t cmd;
-
-  cmd.i = 0;
-  cmd.header.cmd_type = RADEON_CMD_PACKET3;
-
-  return (uint32_t)cmd.i;
-
-}
-#define OUT_BATCH_PACKET3(packet, num_extra) do {	      \
+#define OUT_BATCH_PACKET3_CLIP(packet, num_extra) do {	      \
     if (!b_l_rmesa->radeonScreen->kernel_mm) {		      \
-      OUT_BATCH(cmdpacket3(0));			      \
-      OUT_BATCH(packet);				      \
+      OUT_BATCH(cmdpacket3(RADEON_CMD_PACKET3_CLIP));	      \
+      OUT_BATCH(CP_PACKET3((packet), (num_extra)));	      \
     } else {						      \
       OUT_BATCH(CP_PACKET2);				      \
       OUT_BATCH(CP_PACKET3((packet), (num_extra)));	      \
diff --git a/src/mesa/drivers/dri/r200/r200_maos_arrays.c b/src/mesa/drivers/dri/r200/r200_maos_arrays.c
index c3c2b973b5..7f7ae2ba79 100644
--- a/src/mesa/drivers/dri/r200/r200_maos_arrays.c
+++ b/src/mesa/drivers/dri/r200/r200_maos_arrays.c
@@ -223,6 +223,7 @@ void r200EmitArrays( GLcontext *ctx, GLubyte *vimap_rev )
    GLuint count = VB->Count;
    GLuint i, emitsize;
 
+   fprintf(stderr,"emit arrays\n");
    for ( i = 0; i < 15; i++ ) {
       GLubyte attrib = vimap_rev[i];
       if (attrib != 255) {
@@ -257,14 +258,14 @@ void r200EmitArrays( GLcontext *ctx, GLubyte *vimap_rev )
 	    if (!rmesa->tcl.vertex_data[i].buf) {
 	       if (ctx->VertexProgram._Enabled)
 		  rcommon_emit_vector( ctx,
-				       &(rmesa->tcl.aos[i]),
+				       &(rmesa->tcl.aos[nr]),
 				       (char *)VB->AttribPtr[attrib]->data,
 				       1,
 				       VB->AttribPtr[attrib]->stride,
 				       count);
 	       else
 		 r200_emit_vecfog( ctx,
-				   &(rmesa->tcl.aos[i]),
+				   &(rmesa->tcl.aos[nr]),
 				   (char *)VB->AttribPtr[attrib]->data,
 				   VB->AttribPtr[attrib]->stride,
 				   count);
@@ -313,7 +314,7 @@ void r200EmitArrays( GLcontext *ctx, GLubyte *vimap_rev )
 	 }
 	 if (!rmesa->tcl.vertex_data[i].buf) {
 	   rcommon_emit_vector( ctx,
-				&(rmesa->tcl.aos[i]),
+				&(rmesa->tcl.aos[nr]),
 				(char *)VB->AttribPtr[attrib]->data,
 				emitsize,
 				VB->AttribPtr[attrib]->stride,
@@ -321,6 +322,7 @@ void r200EmitArrays( GLcontext *ctx, GLubyte *vimap_rev )
 	 }
 after_emit:
 	 assert(nr < 12);
+	 nr++;
 	 //	 component[nr++] = &rmesa->tcl.vertex_data[i];
       }
    }
@@ -339,12 +341,10 @@ after_emit:
 void r200ReleaseArrays( GLcontext *ctx, GLuint newinputs )
 {
    r200ContextPtr rmesa = R200_CONTEXT( ctx );
-
-   /* only do it for changed inputs ? */
    int i;
-   for (i = 0; i < 15; i++) {
-     //      if (newinputs & (1 << i))
-	//	 r200ReleaseDmaRegion( rmesa,
-	//	    &rmesa->tcl.vertex_data[i], __FUNCTION__ );
+   for (i = 0; i < rmesa->tcl.nr_aos_components; i++) {
+     if (rmesa->tcl.aos[i].bo) {
+       rmesa->tcl.aos[i].bo = radeon_bo_unref(rmesa->tcl.aos[i].bo);
+     }
    }
 }
diff --git a/src/mesa/drivers/dri/r200/r200_state_init.c b/src/mesa/drivers/dri/r200/r200_state_init.c
index f46b01b56a..3b56fec78a 100644
--- a/src/mesa/drivers/dri/r200/r200_state_init.c
+++ b/src/mesa/drivers/dri/r200/r200_state_init.c
@@ -43,6 +43,10 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "tnl/t_pipeline.h"
 #include "swrast_setup/swrast_setup.h"
 
+#include "radeon_buffer.h"
+#include "radeon_cs.h"
+#include "common_context.h"
+#include "common_cmdbuf.h"
 #include "r200_context.h"
 #include "r200_ioctl.h"
 #include "r200_state.h"
@@ -288,32 +292,66 @@ VP_CHECK( tcl_vp_size, ctx->VertexProgram.Current->Base.NumNativeInstructions >
 VP_CHECK( tcl_vpp_size, ctx->VertexProgram.Current->Base.NumNativeParameters > 96 )
 
 
-#if 0
-static int ctx_emit(GLcontext *ctx, struct radeon_state_atom *atom)
+static void ctx_emit(GLcontext *ctx, struct radeon_state_atom *atom)
 {
    r200ContextPtr r200 = R200_CONTEXT(ctx);
    BATCH_LOCALS(&r200->radeon);
    struct radeon_renderbuffer *rrb;
    uint32_t cbpitch;
+   uint32_t zbpitch;
+   uint32_t dwords = atom->cmd_size;
    GLframebuffer *fb = r200->radeon.dri.drawable->driverPrivate;
-   
+
+   /* output the first 7 bytes of context */
+   BEGIN_BATCH_NO_AUTOSTATE(dwords);
+   OUT_BATCH_TABLE(atom->cmd, 5);
+
+   rrb = r200->radeon.state.depth.rrb;
+   if (!rrb) {
+     OUT_BATCH(atom->cmd[CTX_RB3D_DEPTHOFFSET]);
+     OUT_BATCH(atom->cmd[CTX_RB3D_DEPTHPITCH]);
+   } else {
+     zbpitch = (rrb->pitch / rrb->cpp);
+     OUT_BATCH_RELOC(0, rrb->bo, 0, 0, RADEON_GEM_DOMAIN_VRAM, 0);
+     OUT_BATCH(zbpitch);
+   }
+     
+   OUT_BATCH(atom->cmd[CTX_RB3D_ZSTENCILCNTL]);
+   OUT_BATCH(atom->cmd[CTX_CMD_1]);
+   OUT_BATCH(atom->cmd[CTX_PP_CNTL]);
+   OUT_BATCH(atom->cmd[CTX_RB3D_CNTL]);
+
    rrb = r200->radeon.state.color.rrb;
    if (r200->radeon.radeonScreen->driScreen->dri2.enabled) {
       rrb = (struct radeon_renderbuffer *)fb->Attachment[BUFFER_BACK_LEFT].Renderbuffer;
    }
    if (!rrb || !rrb->bo) {
-      fprintf(stderr, "no rrb\n");
-      return;
+     OUT_BATCH(atom->cmd[CTX_RB3D_COLOROFFSET]);
+   } else {
+     OUT_BATCH_RELOC(0, rrb->bo, 0, 0, RADEON_GEM_DOMAIN_VRAM, 0);
    }
 
-   cbpitch = (rrb->pitch / rrb->cpp);
-   if (rrb->cpp == 4)
-      ;
-   else
-      ;
+   OUT_BATCH(atom->cmd[CTX_CMD_2]);
+
+   if (!rrb || !rrb->bo) {
+     OUT_BATCH(atom->cmd[CTX_RB3D_COLORPITCH]);
+   } else {
+     cbpitch = (rrb->pitch / rrb->cpp);
+     if (rrb->cpp == 4)
+       ;
+     else
+       ;
+     if (r200->radeon.sarea->tiling_enabled)
+       cbpitch |= R200_COLOR_TILE_ENABLE;
+     OUT_BATCH(cbpitch);
+   }
+
+   if (atom->cmd_size == CTX_STATE_SIZE_NEWDRM)
+     OUT_BATCH_TABLE((atom->cmd + 14), 4);
+
+   END_BATCH();
    
 }
-#endif
 
 static int tex_emit(GLcontext *ctx, struct radeon_state_atom *atom)
 {
@@ -410,7 +448,7 @@ void r200InitState( r200ContextPtr rmesa )
    else
       ALLOC_STATE( ctx, always, CTX_STATE_SIZE_OLDDRM, "CTX/context", 0 );
 
-   //   rmesa->hw.ctx.emit = ctx_emit;
+   rmesa->hw.ctx.emit = ctx_emit;
    ALLOC_STATE( set, always, SET_STATE_SIZE, "SET/setup", 0 );
    ALLOC_STATE( lin, always, LIN_STATE_SIZE, "LIN/line", 0 );
    ALLOC_STATE( msk, always, MSK_STATE_SIZE, "MSK/mask", 0 );
diff --git a/src/mesa/drivers/dri/r200/r200_tcl.c b/src/mesa/drivers/dri/r200/r200_tcl.c
index 4a31578c89..226bf1620c 100644
--- a/src/mesa/drivers/dri/r200/r200_tcl.c
+++ b/src/mesa/drivers/dri/r200/r200_tcl.c
@@ -142,6 +142,7 @@ static GLboolean discrete_prim[0x10] = {
 
 static GLushort *r200AllocElts( r200ContextPtr rmesa, GLuint nr ) 
 {
+   fprintf(stderr,"alloc elts\n");
    if (rmesa->dma.flush == r200FlushElts &&
        rmesa->store.cmd_used + nr*2 < R200_CMD_BUF_SZ) {
 
@@ -187,6 +188,7 @@ static void r200EmitPrim( GLcontext *ctx,
    r200ContextPtr rmesa = R200_CONTEXT( ctx );
    r200TclPrimitive( ctx, prim, hwprim );
    
+   fprintf(stderr,"Emit prim %d\n", rmesa->tcl.nr_aos_components);
    rcommonEnsureCmdBufSpace( rmesa, AOS_BUFSZ(rmesa->tcl.nr_aos_components) +
 			     rmesa->hw.max_state_size + VBUF_BUFSZ );