Redoing the way we handle vertex shaders for the draw module.

2 files changed, 59 insertions, 58 deletions

diff --git a/src/mesa/x86/rtasm/x86sse.c b/src/mesa/x86/rtasm/x86sse.c
index a6cfa40f91..f9a9352e46 100644
--- a/src/mesa/x86/rtasm/x86sse.c
+++ b/src/mesa/x86/rtasm/x86sse.c
@@ -6,22 +6,22 @@
 #define DISASSEM 0
 #define X86_TWOB 0x0f
 
-static GLubyte *cptr( void (*label)() )
+static unsigned char *cptr( void (*label)() )
 {
-   return (char *)(unsigned long)label;
+   return (unsigned char *)(unsigned long)label;
 }
 
 
 /* Emit bytes to the instruction stream:
  */
-static void emit_1b( struct x86_function *p, GLbyte b0 )
+static void emit_1b( struct x86_function *p, char b0 )
 {
-   *(GLbyte *)(p->csr++) = b0;
+   *(char *)(p->csr++) = b0;
 }
 
-static void emit_1i( struct x86_function *p, GLint i0 )
+static void emit_1i( struct x86_function *p, int i0 )
 {
-   *(GLint *)(p->csr) = i0;
+   *(int *)(p->csr) = i0;
    p->csr += 4;
 }
 
@@ -35,20 +35,20 @@ static void disassem( struct x86_function *p, const char *fn )
 #endif
 }
 
-static void emit_1ub_fn( struct x86_function *p, GLubyte b0, const char *fn )
+static void emit_1ub_fn( struct x86_function *p, unsigned char b0, const char *fn )
 {
    disassem(p, fn);
    *(p->csr++) = b0;
 }
 
-static void emit_2ub_fn( struct x86_function *p, GLubyte b0, GLubyte b1, const char *fn )
+static void emit_2ub_fn( struct x86_function *p, unsigned char b0, unsigned char b1, const char *fn )
 {
    disassem(p, fn);
    *(p->csr++) = b0;
    *(p->csr++) = b1;
 }
 
-static void emit_3ub_fn( struct x86_function *p, GLubyte b0, GLubyte b1, GLubyte b2, const char *fn )
+static void emit_3ub_fn( struct x86_function *p, unsigned char b0, unsigned char b1, unsigned char b2, const char *fn )
 {
    disassem(p, fn);
    *(p->csr++) = b0;
@@ -69,7 +69,7 @@ static void emit_modrm( struct x86_function *p,
 			struct x86_reg reg, 
 			struct x86_reg regmem )
 {
-   GLubyte val = 0;
+   unsigned char val = 0;
    
    assert(reg.mod == mod_REG);
    
@@ -104,7 +104,7 @@ static void emit_modrm( struct x86_function *p,
 
 
 static void emit_modrm_noreg( struct x86_function *p,
-			      GLuint op,
+			      unsigned op,
 			      struct x86_reg regmem )
 {
    struct x86_reg dummy = x86_make_reg(file_REG32, op);
@@ -117,8 +117,8 @@ static void emit_modrm_noreg( struct x86_function *p,
  * the arguments presented.
  */
 static void emit_op_modrm( struct x86_function *p,
-			   GLubyte op_dst_is_reg, 
-			   GLubyte op_dst_is_mem,
+			   unsigned char op_dst_is_reg, 
+			   unsigned char op_dst_is_mem,
 			   struct x86_reg dst,
 			   struct x86_reg src )
 {  
@@ -162,7 +162,7 @@ struct x86_reg x86_make_reg( enum x86_reg_file file,
 }
 
 struct x86_reg x86_make_disp( struct x86_reg reg,
-			      GLint disp )
+			      int disp )
 {
    assert(reg.file == file_REG32);
 
@@ -191,7 +191,7 @@ struct x86_reg x86_get_base_reg( struct x86_reg reg )
    return x86_make_reg( reg.file, reg.idx );
 }
 
-GLubyte *x86_get_label( struct x86_function *p )
+unsigned char *x86_get_label( struct x86_function *p )
 {
    return p->csr;
 }
@@ -205,13 +205,13 @@ GLubyte *x86_get_label( struct x86_function *p )
 
 void x86_jcc( struct x86_function *p,
 	      enum x86_cc cc,
-	      GLubyte *label )
+	      unsigned char *label )
 {
-   GLint offset = label - (x86_get_label(p) + 2);
+   int offset = label - (x86_get_label(p) + 2);
    
    if (offset <= 127 && offset >= -128) {
       emit_1ub(p, 0x70 + cc);
-      emit_1b(p, (GLbyte) offset);
+      emit_1b(p, (char) offset);
    }
    else {
       offset = label - (x86_get_label(p) + 6);
@@ -222,7 +222,7 @@ void x86_jcc( struct x86_function *p,
 
 /* Always use a 32bit offset for forward jumps:
  */
-GLubyte *x86_jcc_forward( struct x86_function *p,
+unsigned char *x86_jcc_forward( struct x86_function *p,
 			  enum x86_cc cc )
 {
    emit_2ub(p, 0x0f, 0x80 + cc);
@@ -230,14 +230,14 @@ GLubyte *x86_jcc_forward( struct x86_function *p,
    return x86_get_label(p);
 }
 
-GLubyte *x86_jmp_forward( struct x86_function *p)
+unsigned char *x86_jmp_forward( struct x86_function *p)
 {
    emit_1ub(p, 0xe9);
    emit_1i(p, 0);
    return x86_get_label(p);
 }
 
-GLubyte *x86_call_forward( struct x86_function *p)
+unsigned char *x86_call_forward( struct x86_function *p)
 {
    emit_1ub(p, 0xe8);
    emit_1i(p, 0);
@@ -247,12 +247,12 @@ GLubyte *x86_call_forward( struct x86_function *p)
 /* Fixup offset from forward jump:
  */
 void x86_fixup_fwd_jump( struct x86_function *p,
-			 GLubyte *fixup )
+			 unsigned char *fixup )
 {
    *(int *)(fixup - 4) = x86_get_label(p) - fixup;
 }
 
-void x86_jmp( struct x86_function *p, GLubyte *label)
+void x86_jmp( struct x86_function *p, unsigned char *label)
 {
    emit_1ub(p, 0xe9);
    emit_1i(p, label - x86_get_label(p) - 4);
@@ -268,7 +268,7 @@ void x86_call( struct x86_function *p, void (*label)())
  * Temporary. As I need immediate operands, and dont want to mess with the codegen,
  * I load the immediate into general purpose register and use it.
  */
-void x86_mov_reg_imm( struct x86_function *p, struct x86_reg dst, GLint imm )
+void x86_mov_reg_imm( struct x86_function *p, struct x86_reg dst, int imm )
 {
    assert(dst.mod == mod_REG);
    emit_1ub(p, 0xb8 + dst.idx);
@@ -595,7 +595,7 @@ void sse_cvtps2pi( struct x86_function *p,
 void sse_shufps( struct x86_function *p,
 		 struct x86_reg dest,
 		 struct x86_reg arg0,
-		 GLubyte shuf) 
+		 unsigned char shuf) 
 {
    emit_2ub(p, X86_TWOB, 0xC6);
    emit_modrm(p, dest, arg0);
@@ -605,7 +605,7 @@ void sse_shufps( struct x86_function *p,
 void sse_cmpps( struct x86_function *p,
 		struct x86_reg dest,
 		struct x86_reg arg0,
-		GLubyte cc) 
+		unsigned char cc) 
 {
    emit_2ub(p, X86_TWOB, 0xC2);
    emit_modrm(p, dest, arg0);
@@ -630,7 +630,7 @@ void sse_pmovmskb( struct x86_function *p,
 void sse2_pshufd( struct x86_function *p,
 		  struct x86_reg dest,
 		  struct x86_reg arg0,
-		  GLubyte shuf) 
+		  unsigned char shuf) 
 {
    emit_3ub(p, 0x66, X86_TWOB, 0x70);
    emit_modrm(p, dest, arg0);
@@ -772,11 +772,11 @@ void x87_fclex( struct x86_function *p )
 
 
 static void x87_arith_op( struct x86_function *p, struct x86_reg dst, struct x86_reg arg,
-			  GLubyte dst0ub0,
-			  GLubyte dst0ub1,
-			  GLubyte arg0ub0,
-			  GLubyte arg0ub1,
-			  GLubyte argmem_noreg)
+			  unsigned char dst0ub0,
+			  unsigned char dst0ub1,
+			  unsigned char arg0ub0,
+			  unsigned char arg0ub1,
+			  unsigned char argmem_noreg)
 {
    assert(dst.file == file_x87);
 
@@ -1116,7 +1116,7 @@ void mmx_movq( struct x86_function *p,
  * account any push/pop activity:
  */
 struct x86_reg x86_fn_arg( struct x86_function *p,
-			   GLuint arg )
+			   unsigned arg )
 {
    return x86_make_disp(x86_make_reg(file_REG32, reg_SP), 
 			p->stack_offset + arg * 4);	/* ??? */
@@ -1128,7 +1128,7 @@ void x86_init_func( struct x86_function *p )
    x86_init_func_size(p, 1024);
 }
 
-void x86_init_func_size( struct x86_function *p, GLuint code_size )
+void x86_init_func_size( struct x86_function *p, unsigned code_size )
 {
    p->store = _mesa_exec_malloc(code_size);
    p->csr = p->store;
diff --git a/src/mesa/x86/rtasm/x86sse.h b/src/mesa/x86/rtasm/x86sse.h
index c1ca06088b..63b9a36392 100644
--- a/src/mesa/x86/rtasm/x86sse.h
+++ b/src/mesa/x86/rtasm/x86sse.h
@@ -4,24 +4,22 @@
 
 #if defined(__i386__) || defined(__386__)
 
-#include "glheader.h"
-
 /* It is up to the caller to ensure that instructions issued are
  * suitable for the host cpu.  There are no checks made in this module
  * for mmx/sse/sse2 support on the cpu.
  */
 struct x86_reg {
-   GLuint file:3;
-   GLuint idx:3;
-   GLuint mod:2;		/* mod_REG if this is just a register */
-   GLint  disp:24;		/* only +/- 23bits of offset - should be enough... */
+   unsigned file:3;
+   unsigned idx:3;
+   unsigned mod:2;		/* mod_REG if this is just a register */
+   int      disp:24;		/* only +/- 23bits of offset - should be enough... */
 };
 
 struct x86_function {
-   GLubyte *store;
-   GLubyte *csr;
-   GLuint stack_offset;
-   GLint need_emms;
+   unsigned char *store;
+   unsigned char *csr;
+   unsigned stack_offset;
+   int need_emms;
    const char *fn;
 };
 
@@ -81,7 +79,7 @@ enum sse_cc {
 
 
 void x86_init_func( struct x86_function *p );
-void x86_init_func_size( struct x86_function *p, GLuint code_size );
+void x86_init_func_size( struct x86_function *p, unsigned code_size );
 void x86_release_func( struct x86_function *p );
 void (*x86_get_func( struct x86_function *p ))( void );
 
@@ -93,7 +91,7 @@ struct x86_reg x86_make_reg( enum x86_reg_file file,
 			     enum x86_reg_name idx );
 
 struct x86_reg x86_make_disp( struct x86_reg reg,
-			      GLint disp );
+			      int disp );
 
 struct x86_reg x86_deref( struct x86_reg reg );
 
@@ -102,23 +100,23 @@ struct x86_reg x86_get_base_reg( struct x86_reg reg );
 
 /* Labels, jumps and fixup:
  */
-GLubyte *x86_get_label( struct x86_function *p );
+unsigned char *x86_get_label( struct x86_function *p );
 
 void x86_jcc( struct x86_function *p,
 	      enum x86_cc cc,
-	      GLubyte *label );
+	      unsigned char *label );
 
-GLubyte *x86_jcc_forward( struct x86_function *p,
+unsigned char *x86_jcc_forward( struct x86_function *p,
 			  enum x86_cc cc );
 
-GLubyte *x86_jmp_forward( struct x86_function *p);
+unsigned char *x86_jmp_forward( struct x86_function *p);
 
-GLubyte *x86_call_forward( struct x86_function *p);
+unsigned char *x86_call_forward( struct x86_function *p);
 
 void x86_fixup_fwd_jump( struct x86_function *p,
-			 GLubyte *fixup );
+			 unsigned char *fixup );
 
-void x86_jmp( struct x86_function *p, GLubyte *label );
+void x86_jmp( struct x86_function *p, unsigned char *label );
 
 void x86_call( struct x86_function *p, void (*label)() );
 
@@ -126,7 +124,7 @@ void x86_call( struct x86_function *p, void (*label)() );
  * Temporary. As I need immediate operands, and dont want to mess with the codegen,
  * I load the immediate into general purpose register and use it.
  */
-void x86_mov_reg_imm( struct x86_function *p, struct x86_reg dst, GLint imm );
+void x86_mov_reg_imm( struct x86_function *p, struct x86_reg dst, int imm );
 
 
 /* Macro for sse_shufps() and sse2_pshufd():
@@ -147,7 +145,8 @@ void sse2_movd( struct x86_function *p, struct x86_reg dst, struct x86_reg src )
 void sse2_packssdw( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
 void sse2_packsswb( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
 void sse2_packuswb( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void sse2_pshufd( struct x86_function *p, struct x86_reg dest, struct x86_reg arg0, GLubyte shuf );
+void sse2_pshufd( struct x86_function *p, struct x86_reg dest, struct x86_reg arg0,
+                  unsigned char shuf );
 void sse2_rcpps( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
 void sse2_rcpss( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
 
@@ -157,7 +156,8 @@ void sse_cvtps2pi( struct x86_function *p, struct x86_reg dst, struct x86_reg sr
 void sse_divss( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
 void sse_andnps( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
 void sse_andps( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void sse_cmpps( struct x86_function *p, struct x86_reg dst, struct x86_reg src, GLubyte cc );
+void sse_cmpps( struct x86_function *p, struct x86_reg dst, struct x86_reg src,
+                unsigned char cc );
 void sse_maxps( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
 void sse_maxss( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
 void sse_minps( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
@@ -175,7 +175,8 @@ void sse_xorps( struct x86_function *p, struct x86_reg dst, struct x86_reg src )
 void sse_subps( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
 void sse_rsqrtps( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
 void sse_rsqrtss( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void sse_shufps( struct x86_function *p, struct x86_reg dest, struct x86_reg arg0, GLubyte shuf );
+void sse_shufps( struct x86_function *p, struct x86_reg dest, struct x86_reg arg0,
+                 unsigned char shuf );
 void sse_pmovmskb( struct x86_function *p, struct x86_reg dest, struct x86_reg src );
 
 void x86_add( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
@@ -247,7 +248,7 @@ void x87_fucom( struct x86_function *p, struct x86_reg arg );
  * account any push/pop activity.  Note - doesn't track explict
  * manipulation of ESP by other instructions.
  */
-struct x86_reg x86_fn_arg( struct x86_function *p, GLuint arg );
+struct x86_reg x86_fn_arg( struct x86_function *p, unsigned arg );
 
 #endif
 #endif