diff options
44 files changed, 7645 insertions, 7867 deletions
diff --git a/libX11/modules/im/ximcp/imLcFlt.c b/libX11/modules/im/ximcp/imLcFlt.c index 014ad6add..06aa9980a 100644 --- a/libX11/modules/im/ximcp/imLcFlt.c +++ b/libX11/modules/im/ximcp/imLcFlt.c @@ -1,115 +1,125 @@ -/******************************************************************
-
- Copyright 1992 by Fuji Xerox Co., Ltd.
- Copyright 1992, 1994 by FUJITSU LIMITED
-
-Permission to use, copy, modify, distribute, and sell this software
-and its documentation for any purpose is hereby granted without fee,
-provided that the above copyright notice appear in all copies and
-that both that copyright notice and this permission notice appear
-in supporting documentation, and that the name of Fuji Xerox,
-FUJITSU LIMITED not be used in advertising or publicity pertaining
-to distribution of the software without specific, written prior
-permission. Fuji Xerox, FUJITSU LIMITED make no representations
-about the suitability of this software for any purpose.
-It is provided "as is" without express or implied warranty.
-
-FUJI XEROX, FUJITSU LIMITED DISCLAIM ALL WARRANTIES WITH
-REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF
-MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL FUJI XEROX,
-FUJITSU LIMITED BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL
-DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA
-OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER
-TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR
-PERFORMANCE OF THIS SOFTWARE.
-
- Author : Kazunori Nishihara Fuji Xerox
- Modifier : Takashi Fujiwara FUJITSU LIMITED
- fujiwara@a80.tech.yk.fujitsu.co.jp
-
-******************************************************************/
-
-#ifdef HAVE_CONFIG_H
-#include <config.h>
-#endif
-#include "Xlibint.h"
-#include <X11/keysym.h>
-#include "Xlcint.h"
-#include "Ximint.h"
-
-Bool
-_XimLocalFilter(Display *d, Window w, XEvent *ev, XPointer client_data)
-{
- Xic ic = (Xic)client_data;
- KeySym keysym;
- static char buf[256];
- DefTree *b = ic->private.local.base.tree;
- DTIndex t;
-
- if(ev->xkey.keycode == 0)
- return (False);
-
- XLookupString((XKeyEvent *)ev, buf, sizeof(buf), &keysym, NULL);
-
- if(IsModifierKey(keysym))
- return (False);
-
- if(keysym >= XK_braille_dot_1 && keysym <= XK_braille_dot_8) {
- if(ev->type == KeyPress) {
- ic->private.local.brl_pressed |=
- 1<<(keysym-XK_braille_dot_1);
- } else {
- if(!ic->private.local.brl_committing
- || ev->xkey.time - ic->private.local.brl_release_start > 300) {
- ic->private.local.brl_committing = ic->private.local.brl_pressed;
- ic->private.local.brl_release_start = ev->xkey.time;
- }
- ic->private.local.brl_pressed &= ~(1<<(keysym-XK_braille_dot_1));
- if(!ic->private.local.brl_pressed) {
- if(ic->private.local.brl_committing) {
- ic->private.local.brl_committed =
- ic->private.local.brl_committing;
- ic->private.local.composed = 0;
- ev->type = KeyPress;
- ev->xkey.keycode = 0;
- _XPutBackEvent(d, ev);
- }
- }
- }
- return(True);
- }
-
- if( (ev->type != KeyPress)
- || (((Xim)ic->core.im)->private.local.top == 0 ) )
- return(False);
-
- for(t = ic->private.local.context; t; t = b[t].next) {
- if(((ev->xkey.state & b[t].modifier_mask) == b[t].modifier) &&
- (keysym == b[t].keysym))
- break;
- }
-
- if(t) { /* Matched */
- if(b[t].succession) { /* Intermediate */
- ic->private.local.context = b[t].succession;
- return(True);
- } else { /* Terminate (reached to leaf) */
- ic->private.local.composed = t;
- ic->private.local.brl_committed = 0;
- /* return back to client KeyPressEvent keycode == 0 */
- ev->xkey.keycode = 0;
- XPutBackEvent(d, ev);
- /* initialize internal state for next key sequence */
- ic->private.local.context = ((Xim)ic->core.im)->private.local.top;
- return(True);
- }
- } else { /* Unmatched */
- if(ic->private.local.context == ((Xim)ic->core.im)->private.local.top) {
- return(False);
- }
- /* Error (Sequence Unmatch occured) */
- /* initialize internal state for next key sequence */
- ic->private.local.context = ((Xim)ic->core.im)->private.local.top;
- return(True);
- }
-}
+/****************************************************************** + + Copyright 1992 by Fuji Xerox Co., Ltd. + Copyright 1992, 1994 by FUJITSU LIMITED + +Permission to use, copy, modify, distribute, and sell this software +and its documentation for any purpose is hereby granted without fee, +provided that the above copyright notice appear in all copies and +that both that copyright notice and this permission notice appear +in supporting documentation, and that the name of Fuji Xerox, +FUJITSU LIMITED not be used in advertising or publicity pertaining +to distribution of the software without specific, written prior +permission. Fuji Xerox, FUJITSU LIMITED make no representations +about the suitability of this software for any purpose. +It is provided "as is" without express or implied warranty. + +FUJI XEROX, FUJITSU LIMITED DISCLAIM ALL WARRANTIES WITH +REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF +MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL FUJI XEROX, +FUJITSU LIMITED BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL +DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA +OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER +TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR +PERFORMANCE OF THIS SOFTWARE. + + Author : Kazunori Nishihara Fuji Xerox + Modifier : Takashi Fujiwara FUJITSU LIMITED + fujiwara@a80.tech.yk.fujitsu.co.jp + +******************************************************************/ + +#ifdef HAVE_CONFIG_H +#include <config.h> +#endif +#include "Xlibint.h" +#include <X11/keysym.h> +#include "Xlcint.h" +#include "Ximint.h" + +Bool +_XimLocalFilter(Display *d, Window w, XEvent *ev, XPointer client_data) +{ + Xic ic = (Xic)client_data; + KeySym keysym; + static char buf[256]; + DefTree *b = ic->private.local.base.tree; + DTIndex t; + Bool braille = False; + + if(ev->xkey.keycode == 0) + return (False); + + XLookupString((XKeyEvent *)ev, buf, sizeof(buf), &keysym, NULL); + + if(IsModifierKey(keysym)) + return (False); + + if(keysym >= XK_braille_dot_1 && keysym <= XK_braille_dot_8) { + if(ev->type == KeyPress) { + ic->private.local.brl_pressed |= + 1<<(keysym-XK_braille_dot_1); + return(True); + } else { + if(!ic->private.local.brl_committing + || ev->xkey.time - ic->private.local.brl_release_start > 300) { + ic->private.local.brl_committing = ic->private.local.brl_pressed; + ic->private.local.brl_release_start = ev->xkey.time; + } + ic->private.local.brl_pressed &= ~(1<<(keysym-XK_braille_dot_1)); + if(!ic->private.local.brl_pressed && ic->private.local.brl_committing) { + /* Commited a braille pattern, let it go through compose tree */ + keysym = XK_braille_blank | ic->private.local.brl_committing; + ev->type = KeyPress; + braille = True; + } else { + return(True); + } + } + } + + if( (ev->type != KeyPress) + || (((Xim)ic->core.im)->private.local.top == 0 ) ) + goto emit_braille; + + for(t = ic->private.local.context; t; t = b[t].next) { + if(((ev->xkey.state & b[t].modifier_mask) == b[t].modifier) && + (keysym == b[t].keysym)) + break; + } + + if(t) { /* Matched */ + if(b[t].succession) { /* Intermediate */ + ic->private.local.context = b[t].succession; + return(True); + } else { /* Terminate (reached to leaf) */ + ic->private.local.composed = t; + ic->private.local.brl_committed = 0; + /* return back to client KeyPressEvent keycode == 0 */ + ev->xkey.keycode = 0; + XPutBackEvent(d, ev); + /* initialize internal state for next key sequence */ + ic->private.local.context = ((Xim)ic->core.im)->private.local.top; + return(True); + } + } else { /* Unmatched */ + if(ic->private.local.context == ((Xim)ic->core.im)->private.local.top) { + goto emit_braille; + } + /* Error (Sequence Unmatch occured) */ + /* initialize internal state for next key sequence */ + ic->private.local.context = ((Xim)ic->core.im)->private.local.top; + return(True); + } + +emit_braille: + if(braille) { + /* Braille pattern is not in compose tree, emit alone */ + ic->private.local.brl_committed = ic->private.local.brl_committing; + ic->private.local.composed = 0; + ev->xkey.keycode = 0; + _XPutBackEvent(d, ev); + return(True); + } + return(False); +} diff --git a/libX11/modules/im/ximcp/imLcIc.c b/libX11/modules/im/ximcp/imLcIc.c index 53d53efec..c0728084f 100644 --- a/libX11/modules/im/ximcp/imLcIc.c +++ b/libX11/modules/im/ximcp/imLcIc.c @@ -180,7 +180,7 @@ _XimLocalCreateIC( values, XIM_CREATEIC, True)) { goto Set_Error; } - ic_values.filter_events = KeyPressMask; + ic_values.filter_events = KeyPressMask | KeyReleaseMask; _XimSetCurrentICValues(ic, &ic_values); if(_XimSetICDefaults(ic, (XPointer)&ic_values, XIM_SETICDEFAULTS, res, num) == False) { diff --git a/libX11/modules/im/ximcp/imLcLkup.c b/libX11/modules/im/ximcp/imLcLkup.c index 4891176cc..8e4111a6a 100644 --- a/libX11/modules/im/ximcp/imLcLkup.c +++ b/libX11/modules/im/ximcp/imLcLkup.c @@ -63,20 +63,25 @@ _XimLocalMbLookupString(XIC xic, XKeyEvent *ev, char *buffer, int bytes, unsigned char pattern = ic->private.local.brl_committed; char mb[XLC_PUBLIC(ic->core.im->core.lcd, mb_cur_max)]; ret = _Xlcwctomb(ic->core.im->core.lcd, mb, BRL_UC_ROW | pattern); - if(ret < 0) { - if(status) *status = XLookupNone; - return(0); - } if(ret > bytes) { if(status) *status = XBufferOverflow; return(ret); } - if(keysym) { - *keysym = XK_braille_blank | pattern; - if(status) *status = XLookupBoth; - } else - if(status) *status = XLookupChars; - memcpy(buffer, mb, ret); + if(keysym) *keysym = XK_braille_blank | pattern; + if(ret > 0) { + if (keysym) { + if(status) *status = XLookupBoth; + } else { + if(status) *status = XLookupChars; + } + memcpy(buffer, mb, ret); + } else { + if(keysym) { + if(status) *status = XLookupKeySym; + } else { + if(status) *status = XLookupNone; + } + } } else { /* Composed Event */ ret = strlen(&mb[b[ic->private.local.composed].mb]); if(ret > bytes) { @@ -217,6 +222,11 @@ _XimLocalUtf8LookupString(XIC xic, XKeyEvent *ev, char *buffer, int bytes, buffer[0] = 0xe0 | ((BRL_UC_ROW >> 12) & 0x0f); buffer[1] = 0x80 | ((BRL_UC_ROW >> 8) & 0x30) | (pattern >> 6); buffer[2] = 0x80 | (pattern & 0x3f); + if(keysym) { + *keysym = XK_braille_blank | pattern; + if(status) *status = XLookupBoth; + } else + if(status) *status = XLookupChars; } else { /* Composed Event */ ret = strlen(&utf8[b[ic->private.local.composed].utf8]); if(ret > bytes) { diff --git a/mesalib/src/glsl/ast_expr.cpp b/mesalib/src/glsl/ast_expr.cpp index 974beb9f6..e624d11cf 100644 --- a/mesalib/src/glsl/ast_expr.cpp +++ b/mesalib/src/glsl/ast_expr.cpp @@ -20,12 +20,9 @@ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER * DEALINGS IN THE SOFTWARE. */ -#include <cstdio> -#include <cassert> +#include <assert.h> #include "ast.h" -using std::printf; - const char * ast_expression::operator_string(enum ast_operators op) { diff --git a/mesalib/src/glsl/ast_type.cpp b/mesalib/src/glsl/ast_type.cpp index 5ddfeec87..c680ae5f6 100644 --- a/mesalib/src/glsl/ast_type.cpp +++ b/mesalib/src/glsl/ast_type.cpp @@ -21,14 +21,11 @@ * DEALINGS IN THE SOFTWARE. */ -#include <cstdio> #include "ast.h" extern "C" { #include "program/symbol_table.h" } -using std::printf; - void ast_type_specifier::print(void) const { diff --git a/mesalib/src/glsl/glsl_parser_extras.h b/mesalib/src/glsl/glsl_parser_extras.h index 8d4fca787..10cb673c6 100644 --- a/mesalib/src/glsl/glsl_parser_extras.h +++ b/mesalib/src/glsl/glsl_parser_extras.h @@ -31,7 +31,7 @@ #ifdef __cplusplus -#include <cstdlib> +#include <stdlib.h> #include "glsl_symbol_table.h" enum _mesa_glsl_parser_targets { diff --git a/mesalib/src/glsl/glsl_types.cpp b/mesalib/src/glsl/glsl_types.cpp index f4d9242b2..76b4f3e4c 100644 --- a/mesalib/src/glsl/glsl_types.cpp +++ b/mesalib/src/glsl/glsl_types.cpp @@ -21,7 +21,7 @@ * DEALINGS IN THE SOFTWARE. */ -#include <cstdio> +#include <stdio.h> #include <stdlib.h> #include "main/core.h" /* for Elements */ #include "glsl_symbol_table.h" diff --git a/mesalib/src/glsl/glsl_types.h b/mesalib/src/glsl/glsl_types.h index ab6858faa..61bf5e0cf 100644 --- a/mesalib/src/glsl/glsl_types.h +++ b/mesalib/src/glsl/glsl_types.h @@ -26,8 +26,8 @@ #ifndef GLSL_TYPES_H #define GLSL_TYPES_H -#include <cstring> -#include <cassert> +#include <string.h> +#include <assert.h> extern "C" { #include "GL/gl.h" diff --git a/mesalib/src/glsl/hir_field_selection.cpp b/mesalib/src/glsl/hir_field_selection.cpp index 995f284fa..3c33127b5 100644 --- a/mesalib/src/glsl/hir_field_selection.cpp +++ b/mesalib/src/glsl/hir_field_selection.cpp @@ -27,8 +27,6 @@ #include "ast.h" #include "glsl_types.h" -using std::strcmp; - ir_rvalue * _mesa_ast_field_selection_to_hir(const ast_expression *expr, exec_list *instructions, diff --git a/mesalib/src/glsl/ir.h b/mesalib/src/glsl/ir.h index 74a8b06b1..f2f902c0a 100644 --- a/mesalib/src/glsl/ir.h +++ b/mesalib/src/glsl/ir.h @@ -26,8 +26,8 @@ #ifndef IR_H #define IR_H -#include <cstdio> -#include <cstdlib> +#include <stdio.h> +#include <stdlib.h> #include "ralloc.h" #include "glsl_types.h" diff --git a/mesalib/src/glsl/ir_import_prototypes.cpp b/mesalib/src/glsl/ir_import_prototypes.cpp index be5e0c1d3..3585bf6b2 100644 --- a/mesalib/src/glsl/ir_import_prototypes.cpp +++ b/mesalib/src/glsl/ir_import_prototypes.cpp @@ -27,7 +27,6 @@ * * \author Ian Romanick */ -#include <cstdio> #include "ir.h" #include "glsl_symbol_table.h" diff --git a/mesalib/src/glsl/ir_print_visitor.cpp b/mesalib/src/glsl/ir_print_visitor.cpp index be76945a2..82ccc722f 100644 --- a/mesalib/src/glsl/ir_print_visitor.cpp +++ b/mesalib/src/glsl/ir_print_visitor.cpp @@ -25,9 +25,6 @@ #include "glsl_types.h" #include "glsl_parser_extras.h" -using std::printf; -using std::strncmp; - static void print_type(const glsl_type *t); void diff --git a/mesalib/src/glsl/ir_validate.cpp b/mesalib/src/glsl/ir_validate.cpp index b0dd6c21f..44d7549ea 100644 --- a/mesalib/src/glsl/ir_validate.cpp +++ b/mesalib/src/glsl/ir_validate.cpp @@ -39,9 +39,6 @@ #include "program/hash_table.h" #include "glsl_types.h" -using std::abort; -using std::printf; - class ir_validate : public ir_hierarchical_visitor { public: ir_validate() diff --git a/mesalib/src/glsl/link_functions.cpp b/mesalib/src/glsl/link_functions.cpp index 861fa39b5..5851c14d6 100644 --- a/mesalib/src/glsl/link_functions.cpp +++ b/mesalib/src/glsl/link_functions.cpp @@ -21,10 +21,6 @@ * DEALINGS IN THE SOFTWARE. */ -#include <cstdlib> -#include <cstdio> -#include <cstdarg> - #include "main/core.h" #include "glsl_symbol_table.h" #include "glsl_parser_extras.h" diff --git a/mesalib/src/glsl/linker.cpp b/mesalib/src/glsl/linker.cpp index 6c003bb02..7db5c5e8d 100644 --- a/mesalib/src/glsl/linker.cpp +++ b/mesalib/src/glsl/linker.cpp @@ -63,10 +63,6 @@ * * \author Ian Romanick <ian.d.romanick@intel.com> */ -#include <cstdlib> -#include <cstdio> -#include <cstdarg> -#include <climits> #include "main/core.h" #include "glsl_symbol_table.h" diff --git a/mesalib/src/glsl/loop_controls.cpp b/mesalib/src/glsl/loop_controls.cpp index 9eaa50f22..9acbadc50 100644 --- a/mesalib/src/glsl/loop_controls.cpp +++ b/mesalib/src/glsl/loop_controls.cpp @@ -21,7 +21,7 @@ * DEALINGS IN THE SOFTWARE. */ -#include <climits> +#include <limits.h> #include "main/compiler.h" #include "glsl_types.h" #include "loop_analysis.h" diff --git a/mesalib/src/glsl/lower_mat_op_to_vec.cpp b/mesalib/src/glsl/lower_mat_op_to_vec.cpp index bdc53a1f8..8cbbfa713 100644 --- a/mesalib/src/glsl/lower_mat_op_to_vec.cpp +++ b/mesalib/src/glsl/lower_mat_op_to_vec.cpp @@ -35,9 +35,6 @@ #include "ir_expression_flattening.h" #include "glsl_types.h" -using std::abort; -using std::printf; - class ir_mat_op_to_vec_visitor : public ir_hierarchical_visitor { public: ir_mat_op_to_vec_visitor() diff --git a/mesalib/src/glsl/main.cpp b/mesalib/src/glsl/main.cpp index 6f0552931..096da93dc 100644 --- a/mesalib/src/glsl/main.cpp +++ b/mesalib/src/glsl/main.cpp @@ -20,8 +20,6 @@ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER * DEALINGS IN THE SOFTWARE. */ -#include <cstdlib> -#include <cstdio> #include <getopt.h> #include "ast.h" diff --git a/mesalib/src/glsl/opt_constant_propagation.cpp b/mesalib/src/glsl/opt_constant_propagation.cpp index 814f6e61f..4425f4211 100644 --- a/mesalib/src/glsl/opt_constant_propagation.cpp +++ b/mesalib/src/glsl/opt_constant_propagation.cpp @@ -41,8 +41,6 @@ #include "ir_optimization.h" #include "glsl_types.h" -using std::memset; - class acp_entry : public exec_node { public: diff --git a/mesalib/src/glsl/opt_constant_variable.cpp b/mesalib/src/glsl/opt_constant_variable.cpp index 9dca0ff8e..3fa7c3bad 100644 --- a/mesalib/src/glsl/opt_constant_variable.cpp +++ b/mesalib/src/glsl/opt_constant_variable.cpp @@ -37,9 +37,6 @@ #include "ir_optimization.h" #include "glsl_types.h" -using std::calloc; -using std::free; - struct assignment_entry { exec_node link; int assignment_count; diff --git a/mesalib/src/glsl/opt_dead_code.cpp b/mesalib/src/glsl/opt_dead_code.cpp index f47b9613e..cb500d2d1 100644 --- a/mesalib/src/glsl/opt_dead_code.cpp +++ b/mesalib/src/glsl/opt_dead_code.cpp @@ -32,8 +32,6 @@ #include "ir_variable_refcount.h" #include "glsl_types.h" -using std::printf; - static bool debug = false; /** diff --git a/mesalib/src/glsl/opt_dead_code_local.cpp b/mesalib/src/glsl/opt_dead_code_local.cpp index f67b46337..39962bd60 100644 --- a/mesalib/src/glsl/opt_dead_code_local.cpp +++ b/mesalib/src/glsl/opt_dead_code_local.cpp @@ -38,8 +38,6 @@ #include "ir_optimization.h" #include "glsl_types.h" -using std::printf; - static bool debug = false; class assignment_entry : public exec_node diff --git a/mesalib/src/glsl/opt_dead_functions.cpp b/mesalib/src/glsl/opt_dead_functions.cpp index d72eb6115..ceb79080a 100644 --- a/mesalib/src/glsl/opt_dead_functions.cpp +++ b/mesalib/src/glsl/opt_dead_functions.cpp @@ -32,8 +32,6 @@ #include "ir_expression_flattening.h" #include "glsl_types.h" - using std::strcmp; - class signature_entry : public exec_node { public: diff --git a/mesalib/src/glsl/opt_structure_splitting.cpp b/mesalib/src/glsl/opt_structure_splitting.cpp index 8686da06a..014407c0b 100644 --- a/mesalib/src/glsl/opt_structure_splitting.cpp +++ b/mesalib/src/glsl/opt_structure_splitting.cpp @@ -38,9 +38,6 @@ #include "ir_rvalue_visitor.h" #include "glsl_types.h" -using std::printf; -using std::strcmp; - static bool debug = false; // XXX using variable_entry2 here to avoid collision (MSVC multiply-defined diff --git a/mesalib/src/glsl/opt_swizzle_swizzle.cpp b/mesalib/src/glsl/opt_swizzle_swizzle.cpp index 8d0e1051d..bc442fa86 100644 --- a/mesalib/src/glsl/opt_swizzle_swizzle.cpp +++ b/mesalib/src/glsl/opt_swizzle_swizzle.cpp @@ -32,8 +32,6 @@ #include "ir_optimization.h" #include "glsl_types.h" -using std::memset; - class ir_swizzle_swizzle_visitor : public ir_hierarchical_visitor { public: ir_swizzle_swizzle_visitor() diff --git a/mesalib/src/glsl/opt_tree_grafting.cpp b/mesalib/src/glsl/opt_tree_grafting.cpp index a85ba8234..1ef940f9c 100644 --- a/mesalib/src/glsl/opt_tree_grafting.cpp +++ b/mesalib/src/glsl/opt_tree_grafting.cpp @@ -54,8 +54,6 @@ #include "ir_optimization.h" #include "glsl_types.h" -using std::printf; - static bool debug = false; class ir_tree_grafting_visitor : public ir_hierarchical_visitor { diff --git a/mesalib/src/glsl/s_expression.cpp b/mesalib/src/glsl/s_expression.cpp index 77ac08ac7..a922a50d3 100644 --- a/mesalib/src/glsl/s_expression.cpp +++ b/mesalib/src/glsl/s_expression.cpp @@ -22,9 +22,6 @@ * DEALINGS IN THE SOFTWARE. */ -#include <cstdio> -#include <cstdlib> -#include <cstring> #include <assert.h> #include "s_expression.h" diff --git a/mesalib/src/mesa/main/imports.c b/mesalib/src/mesa/main/imports.c index f98098230..bf89815f2 100644 --- a/mesalib/src/mesa/main/imports.c +++ b/mesalib/src/mesa/main/imports.c @@ -1,1033 +1,1028 @@ -/**
- * \file imports.c
- * Standard C library function wrappers.
- *
- * Imports are services which the device driver or window system or
- * operating system provides to the core renderer. The core renderer (Mesa)
- * will call these functions in order to do memory allocation, simple I/O,
- * etc.
- *
- * Some drivers will want to override/replace this file with something
- * specialized, but that'll be rare.
- *
- * Eventually, I want to move roll the glheader.h file into this.
- *
- * \todo Functions still needed:
- * - scanf
- * - qsort
- * - rand and RAND_MAX
- */
-
-/*
- * Mesa 3-D graphics library
- * Version: 7.1
- *
- * Copyright (C) 1999-2007 Brian Paul All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included
- * in all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
- * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
- * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
- * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- */
-
-
-
-#include "imports.h"
-#include "context.h"
-#include "mtypes.h"
-#include "version.h"
-
-#ifdef _GNU_SOURCE
-#include <locale.h>
-#ifdef __APPLE__
-#include <xlocale.h>
-#endif
-#endif
-
-
-#define MAXSTRING 4000 /* for vsnprintf() */
-
-#ifdef WIN32
-#define vsnprintf _vsnprintf
-#elif defined(__IBMC__) || defined(__IBMCPP__) || ( defined(__VMS) && __CRTL_VER < 70312000 )
-extern int vsnprintf(char *str, size_t count, const char *fmt, va_list arg);
-#ifdef __VMS
-#include "vsnprintf.c"
-#endif
-#endif
-
-/**********************************************************************/
-/** \name Memory */
-/*@{*/
-
-/**
- * Allocate aligned memory.
- *
- * \param bytes number of bytes to allocate.
- * \param alignment alignment (must be greater than zero).
- *
- * Allocates extra memory to accommodate rounding up the address for
- * alignment and to record the real malloc address.
- *
- * \sa _mesa_align_free().
- */
-void *
-_mesa_align_malloc(size_t bytes, unsigned long alignment)
-{
-#if defined(HAVE_POSIX_MEMALIGN)
- void *mem;
- int err = posix_memalign(& mem, alignment, bytes);
- if (err)
- return NULL;
- return mem;
-#elif defined(_WIN32) && defined(_MSC_VER)
- return _aligned_malloc(bytes, alignment);
-#else
- uintptr_t ptr, buf;
-
- ASSERT( alignment > 0 );
-
- ptr = (uintptr_t) malloc(bytes + alignment + sizeof(void *));
- if (!ptr)
- return NULL;
-
- buf = (ptr + alignment + sizeof(void *)) & ~(uintptr_t)(alignment - 1);
- *(uintptr_t *)(buf - sizeof(void *)) = ptr;
-
-#ifdef DEBUG
- /* mark the non-aligned area */
- while ( ptr < buf - sizeof(void *) ) {
- *(unsigned long *)ptr = 0xcdcdcdcd;
- ptr += sizeof(unsigned long);
- }
-#endif
-
- return (void *) buf;
-#endif /* defined(HAVE_POSIX_MEMALIGN) */
-}
-
-/**
- * Same as _mesa_align_malloc(), but using calloc(1, ) instead of
- * malloc()
- */
-void *
-_mesa_align_calloc(size_t bytes, unsigned long alignment)
-{
-#if defined(HAVE_POSIX_MEMALIGN)
- void *mem;
-
- mem = _mesa_align_malloc(bytes, alignment);
- if (mem != NULL) {
- (void) memset(mem, 0, bytes);
- }
-
- return mem;
-#elif defined(_WIN32) && defined(_MSC_VER)
- void *mem;
-
- mem = _aligned_malloc(bytes, alignment);
- if (mem != NULL) {
- (void) memset(mem, 0, bytes);
- }
-
- return mem;
-#else
- uintptr_t ptr, buf;
-
- ASSERT( alignment > 0 );
-
- ptr = (uintptr_t) calloc(1, bytes + alignment + sizeof(void *));
- if (!ptr)
- return NULL;
-
- buf = (ptr + alignment + sizeof(void *)) & ~(uintptr_t)(alignment - 1);
- *(uintptr_t *)(buf - sizeof(void *)) = ptr;
-
-#ifdef DEBUG
- /* mark the non-aligned area */
- while ( ptr < buf - sizeof(void *) ) {
- *(unsigned long *)ptr = 0xcdcdcdcd;
- ptr += sizeof(unsigned long);
- }
-#endif
-
- return (void *)buf;
-#endif /* defined(HAVE_POSIX_MEMALIGN) */
-}
-
-/**
- * Free memory which was allocated with either _mesa_align_malloc()
- * or _mesa_align_calloc().
- * \param ptr pointer to the memory to be freed.
- * The actual address to free is stored in the word immediately before the
- * address the client sees.
- */
-void
-_mesa_align_free(void *ptr)
-{
-#if defined(HAVE_POSIX_MEMALIGN)
- free(ptr);
-#elif defined(_WIN32) && defined(_MSC_VER)
- _aligned_free(ptr);
-#else
- void **cubbyHole = (void **) ((char *) ptr - sizeof(void *));
- void *realAddr = *cubbyHole;
- free(realAddr);
-#endif /* defined(HAVE_POSIX_MEMALIGN) */
-}
-
-/**
- * Reallocate memory, with alignment.
- */
-void *
-_mesa_align_realloc(void *oldBuffer, size_t oldSize, size_t newSize,
- unsigned long alignment)
-{
-#if defined(_WIN32) && defined(_MSC_VER)
- (void) oldSize;
- return _aligned_realloc(oldBuffer, newSize, alignment);
-#else
- const size_t copySize = (oldSize < newSize) ? oldSize : newSize;
- void *newBuf = _mesa_align_malloc(newSize, alignment);
- if (newBuf && oldBuffer && copySize > 0) {
- memcpy(newBuf, oldBuffer, copySize);
- }
- if (oldBuffer)
- _mesa_align_free(oldBuffer);
- return newBuf;
-#endif
-}
-
-
-
-/** Reallocate memory */
-void *
-_mesa_realloc(void *oldBuffer, size_t oldSize, size_t newSize)
-{
- const size_t copySize = (oldSize < newSize) ? oldSize : newSize;
- void *newBuffer = malloc(newSize);
- if (newBuffer && oldBuffer && copySize > 0)
- memcpy(newBuffer, oldBuffer, copySize);
- if (oldBuffer)
- free(oldBuffer);
- return newBuffer;
-}
-
-/**
- * Fill memory with a constant 16bit word.
- * \param dst destination pointer.
- * \param val value.
- * \param n number of words.
- */
-void
-_mesa_memset16( unsigned short *dst, unsigned short val, size_t n )
-{
- while (n-- > 0)
- *dst++ = val;
-}
-
-/*@}*/
-
-
-/**********************************************************************/
-/** \name Math */
-/*@{*/
-
-/** Wrapper around sqrt() */
-double
-_mesa_sqrtd(double x)
-{
- return sqrt(x);
-}
-
-
-/*
- * A High Speed, Low Precision Square Root
- * by Paul Lalonde and Robert Dawson
- * from "Graphics Gems", Academic Press, 1990
- *
- * SPARC implementation of a fast square root by table
- * lookup.
- * SPARC floating point format is as follows:
- *
- * BIT 31 30 23 22 0
- * sign exponent mantissa
- */
-static short sqrttab[0x100]; /* declare table of square roots */
-
-void
-_mesa_init_sqrt_table(void)
-{
-#if defined(USE_IEEE) && !defined(DEBUG)
- unsigned short i;
- fi_type fi; /* to access the bits of a float in C quickly */
- /* we use a union defined in glheader.h */
-
- for(i=0; i<= 0x7f; i++) {
- fi.i = 0;
-
- /*
- * Build a float with the bit pattern i as mantissa
- * and an exponent of 0, stored as 127
- */
-
- fi.i = (i << 16) | (127 << 23);
- fi.f = _mesa_sqrtd(fi.f);
-
- /*
- * Take the square root then strip the first 7 bits of
- * the mantissa into the table
- */
-
- sqrttab[i] = (fi.i & 0x7fffff) >> 16;
-
- /*
- * Repeat the process, this time with an exponent of
- * 1, stored as 128
- */
-
- fi.i = 0;
- fi.i = (i << 16) | (128 << 23);
- fi.f = sqrt(fi.f);
- sqrttab[i+0x80] = (fi.i & 0x7fffff) >> 16;
- }
-#else
- (void) sqrttab; /* silence compiler warnings */
-#endif /*HAVE_FAST_MATH*/
-}
-
-
-/**
- * Single precision square root.
- */
-float
-_mesa_sqrtf( float x )
-{
-#if defined(USE_IEEE) && !defined(DEBUG)
- fi_type num;
- /* to access the bits of a float in C
- * we use a union from glheader.h */
-
- short e; /* the exponent */
- if (x == 0.0F) return 0.0F; /* check for square root of 0 */
- num.f = x;
- e = (num.i >> 23) - 127; /* get the exponent - on a SPARC the */
- /* exponent is stored with 127 added */
- num.i &= 0x7fffff; /* leave only the mantissa */
- if (e & 0x01) num.i |= 0x800000;
- /* the exponent is odd so we have to */
- /* look it up in the second half of */
- /* the lookup table, so we set the */
- /* high bit */
- e >>= 1; /* divide the exponent by two */
- /* note that in C the shift */
- /* operators are sign preserving */
- /* for signed operands */
- /* Do the table lookup, based on the quaternary mantissa,
- * then reconstruct the result back into a float
- */
- num.i = ((sqrttab[num.i >> 16]) << 16) | ((e + 127) << 23);
-
- return num.f;
-#else
- return (float) _mesa_sqrtd((double) x);
-#endif
-}
-
-
-/**
- inv_sqrt - A single precision 1/sqrt routine for IEEE format floats.
- written by Josh Vanderhoof, based on newsgroup posts by James Van Buskirk
- and Vesa Karvonen.
-*/
-float
-_mesa_inv_sqrtf(float n)
-{
-#if defined(USE_IEEE) && !defined(DEBUG)
- float r0, x0, y0;
- float r1, x1, y1;
- float r2, x2, y2;
-#if 0 /* not used, see below -BP */
- float r3, x3, y3;
-#endif
- fi_type u;
- unsigned int magic;
-
- /*
- Exponent part of the magic number -
-
- We want to:
- 1. subtract the bias from the exponent,
- 2. negate it
- 3. divide by two (rounding towards -inf)
- 4. add the bias back
-
- Which is the same as subtracting the exponent from 381 and dividing
- by 2.
-
- floor(-(x - 127) / 2) + 127 = floor((381 - x) / 2)
- */
-
- magic = 381 << 23;
-
- /*
- Significand part of magic number -
-
- With the current magic number, "(magic - u.i) >> 1" will give you:
-
- for 1 <= u.f <= 2: 1.25 - u.f / 4
- for 2 <= u.f <= 4: 1.00 - u.f / 8
-
- This isn't a bad approximation of 1/sqrt. The maximum difference from
- 1/sqrt will be around .06. After three Newton-Raphson iterations, the
- maximum difference is less than 4.5e-8. (Which is actually close
- enough to make the following bias academic...)
-
- To get a better approximation you can add a bias to the magic
- number. For example, if you subtract 1/2 of the maximum difference in
- the first approximation (.03), you will get the following function:
-
- for 1 <= u.f <= 2: 1.22 - u.f / 4
- for 2 <= u.f <= 3.76: 0.97 - u.f / 8
- for 3.76 <= u.f <= 4: 0.72 - u.f / 16
- (The 3.76 to 4 range is where the result is < .5.)
-
- This is the closest possible initial approximation, but with a maximum
- error of 8e-11 after three NR iterations, it is still not perfect. If
- you subtract 0.0332281 instead of .03, the maximum error will be
- 2.5e-11 after three NR iterations, which should be about as close as
- is possible.
-
- for 1 <= u.f <= 2: 1.2167719 - u.f / 4
- for 2 <= u.f <= 3.73: 0.9667719 - u.f / 8
- for 3.73 <= u.f <= 4: 0.7167719 - u.f / 16
-
- */
-
- magic -= (int)(0.0332281 * (1 << 25));
-
- u.f = n;
- u.i = (magic - u.i) >> 1;
-
- /*
- Instead of Newton-Raphson, we use Goldschmidt's algorithm, which
- allows more parallelism. From what I understand, the parallelism
- comes at the cost of less precision, because it lets error
- accumulate across iterations.
- */
- x0 = 1.0f;
- y0 = 0.5f * n;
- r0 = u.f;
-
- x1 = x0 * r0;
- y1 = y0 * r0 * r0;
- r1 = 1.5f - y1;
-
- x2 = x1 * r1;
- y2 = y1 * r1 * r1;
- r2 = 1.5f - y2;
-
-#if 1
- return x2 * r2; /* we can stop here, and be conformant -BP */
-#else
- x3 = x2 * r2;
- y3 = y2 * r2 * r2;
- r3 = 1.5f - y3;
-
- return x3 * r3;
-#endif
-#else
- return (float) (1.0 / sqrt(n));
-#endif
-}
-
-/**
- * Find the first bit set in a word.
- */
-int
-_mesa_ffs(int32_t i)
-{
-#if (defined(_WIN32) ) || defined(__IBMC__) || defined(__IBMCPP__)
- register int bit = 0;
- if (i != 0) {
- if ((i & 0xffff) == 0) {
- bit += 16;
- i >>= 16;
- }
- if ((i & 0xff) == 0) {
- bit += 8;
- i >>= 8;
- }
- if ((i & 0xf) == 0) {
- bit += 4;
- i >>= 4;
- }
- while ((i & 1) == 0) {
- bit++;
- i >>= 1;
- }
- bit++;
- }
- return bit;
-#else
- return ffs(i);
-#endif
-}
-
-
-/**
- * Find position of first bit set in given value.
- * XXX Warning: this function can only be used on 64-bit systems!
- * \return position of least-significant bit set, starting at 1, return zero
- * if no bits set.
- */
-int
-_mesa_ffsll(int64_t val)
-{
-#ifdef ffsll
- return ffsll(val);
-#else
- int bit;
-
- assert(sizeof(val) == 8);
-
- bit = _mesa_ffs((int32_t)val);
- if (bit != 0)
- return bit;
-
- bit = _mesa_ffs((int32_t)(val >> 32));
- if (bit != 0)
- return 32 + bit;
-
- return 0;
-#endif
-}
-
-
-/**
- * Return number of bits set in given GLuint.
- */
-unsigned int
-_mesa_bitcount(unsigned int n)
-{
-#if defined(__GNUC__) && \
- ((_GNUC__ == 3 && __GNUC_MINOR__ >= 4) || __GNUC__ >= 4)
- return __builtin_popcount(n);
-#else
- unsigned int bits;
- for (bits = 0; n > 0; n = n >> 1) {
- bits += (n & 1);
- }
- return bits;
-#endif
-}
-
-
-/**
- * Convert a 4-byte float to a 2-byte half float.
- * Based on code from:
- * http://www.opengl.org/discussion_boards/ubb/Forum3/HTML/008786.html
- */
-GLhalfARB
-_mesa_float_to_half(float val)
-{
- const fi_type fi = {val};
- const int flt_m = fi.i & 0x7fffff;
- const int flt_e = (fi.i >> 23) & 0xff;
- const int flt_s = (fi.i >> 31) & 0x1;
- int s, e, m = 0;
- GLhalfARB result;
-
- /* sign bit */
- s = flt_s;
-
- /* handle special cases */
- if ((flt_e == 0) && (flt_m == 0)) {
- /* zero */
- /* m = 0; - already set */
- e = 0;
- }
- else if ((flt_e == 0) && (flt_m != 0)) {
- /* denorm -- denorm float maps to 0 half */
- /* m = 0; - already set */
- e = 0;
- }
- else if ((flt_e == 0xff) && (flt_m == 0)) {
- /* infinity */
- /* m = 0; - already set */
- e = 31;
- }
- else if ((flt_e == 0xff) && (flt_m != 0)) {
- /* NaN */
- m = 1;
- e = 31;
- }
- else {
- /* regular number */
- const int new_exp = flt_e - 127;
- if (new_exp < -24) {
- /* this maps to 0 */
- /* m = 0; - already set */
- e = 0;
- }
- else if (new_exp < -14) {
- /* this maps to a denorm */
- unsigned int exp_val = (unsigned int) (-14 - new_exp); /* 2^-exp_val*/
- e = 0;
- switch (exp_val) {
- case 0:
- _mesa_warning(NULL,
- "float_to_half: logical error in denorm creation!\n");
- /* m = 0; - already set */
- break;
- case 1: m = 512 + (flt_m >> 14); break;
- case 2: m = 256 + (flt_m >> 15); break;
- case 3: m = 128 + (flt_m >> 16); break;
- case 4: m = 64 + (flt_m >> 17); break;
- case 5: m = 32 + (flt_m >> 18); break;
- case 6: m = 16 + (flt_m >> 19); break;
- case 7: m = 8 + (flt_m >> 20); break;
- case 8: m = 4 + (flt_m >> 21); break;
- case 9: m = 2 + (flt_m >> 22); break;
- case 10: m = 1; break;
- }
- }
- else if (new_exp > 15) {
- /* map this value to infinity */
- /* m = 0; - already set */
- e = 31;
- }
- else {
- /* regular */
- e = new_exp + 15;
- m = flt_m >> 13;
- }
- }
-
- result = (s << 15) | (e << 10) | m;
- return result;
-}
-
-
-/**
- * Convert a 2-byte half float to a 4-byte float.
- * Based on code from:
- * http://www.opengl.org/discussion_boards/ubb/Forum3/HTML/008786.html
- */
-float
-_mesa_half_to_float(GLhalfARB val)
-{
- /* XXX could also use a 64K-entry lookup table */
- const int m = val & 0x3ff;
- const int e = (val >> 10) & 0x1f;
- const int s = (val >> 15) & 0x1;
- int flt_m, flt_e, flt_s;
- fi_type fi;
- float result;
-
- /* sign bit */
- flt_s = s;
-
- /* handle special cases */
- if ((e == 0) && (m == 0)) {
- /* zero */
- flt_m = 0;
- flt_e = 0;
- }
- else if ((e == 0) && (m != 0)) {
- /* denorm -- denorm half will fit in non-denorm single */
- const float half_denorm = 1.0f / 16384.0f; /* 2^-14 */
- float mantissa = ((float) (m)) / 1024.0f;
- float sign = s ? -1.0f : 1.0f;
- return sign * mantissa * half_denorm;
- }
- else if ((e == 31) && (m == 0)) {
- /* infinity */
- flt_e = 0xff;
- flt_m = 0;
- }
- else if ((e == 31) && (m != 0)) {
- /* NaN */
- flt_e = 0xff;
- flt_m = 1;
- }
- else {
- /* regular */
- flt_e = e + 112;
- flt_m = m << 13;
- }
-
- fi.i = (flt_s << 31) | (flt_e << 23) | flt_m;
- result = fi.f;
- return result;
-}
-
-/*@}*/
-
-
-/**********************************************************************/
-/** \name Sort & Search */
-/*@{*/
-
-/**
- * Wrapper for bsearch().
- */
-void *
-_mesa_bsearch( const void *key, const void *base, size_t nmemb, size_t size,
- int (*compar)(const void *, const void *) )
-{
-#if defined(_WIN32_WCE)
- void *mid;
- int cmp;
- while (nmemb) {
- nmemb >>= 1;
- mid = (char *)base + nmemb * size;
- cmp = (*compar)(key, mid);
- if (cmp == 0)
- return mid;
- if (cmp > 0) {
- base = (char *)mid + size;
- --nmemb;
- }
- }
- return NULL;
-#else
- return bsearch(key, base, nmemb, size, compar);
-#endif
-}
-
-/*@}*/
-
-
-/**********************************************************************/
-/** \name Environment vars */
-/*@{*/
-
-/**
- * Wrapper for getenv().
- */
-char *
-_mesa_getenv( const char *var )
-{
-#if defined(_XBOX) || defined(_WIN32_WCE)
- return NULL;
-#else
- return getenv(var);
-#endif
-}
-
-/*@}*/
-
-
-/**********************************************************************/
-/** \name String */
-/*@{*/
-
-/**
- * Implemented using malloc() and strcpy.
- * Note that NULL is handled accordingly.
- */
-char *
-_mesa_strdup( const char *s )
-{
- if (s) {
- size_t l = strlen(s);
- char *s2 = (char *) malloc(l + 1);
- if (s2)
- strcpy(s2, s);
- return s2;
- }
- else {
- return NULL;
- }
-}
-
-/** Wrapper around strtof() */
-float
-_mesa_strtof( const char *s, char **end )
-{
-#if defined(_GNU_SOURCE) && !defined(__CYGWIN__) && !defined(__FreeBSD__)
- static locale_t loc = NULL;
- if (!loc) {
- loc = newlocale(LC_CTYPE_MASK, "C", NULL);
- }
- return strtof_l(s, end, loc);
-#elif defined(_ISOC99_SOURCE) || (defined(_XOPEN_SOURCE) && _XOPEN_SOURCE >= 600)
- return strtof(s, end);
-#else
- return (float)strtod(s, end);
-#endif
-}
-
-/** Compute simple checksum/hash for a string */
-unsigned int
-_mesa_str_checksum(const char *str)
-{
- /* This could probably be much better */
- unsigned int sum, i;
- const char *c;
- sum = i = 1;
- for (c = str; *c; c++, i++)
- sum += *c * (i % 100);
- return sum + i;
-}
-
-
-/*@}*/
-
-
-/** Wrapper around vsnprintf() */
-int
-_mesa_snprintf( char *str, size_t size, const char *fmt, ... )
-{
- int r;
- va_list args;
- va_start( args, fmt );
- r = vsnprintf( str, size, fmt, args );
- va_end( args );
- return r;
-}
-
-
-/**********************************************************************/
-/** \name Diagnostics */
-/*@{*/
-
-static void
-output_if_debug(const char *prefixString, const char *outputString,
- GLboolean newline)
-{
- static int debug = -1;
-
- /* Check the MESA_DEBUG environment variable if it hasn't
- * been checked yet. We only have to check it once...
- */
- if (debug == -1) {
- char *env = _mesa_getenv("MESA_DEBUG");
-
- /* In a debug build, we print warning messages *unless*
- * MESA_DEBUG is 0. In a non-debug build, we don't
- * print warning messages *unless* MESA_DEBUG is
- * set *to any value*.
- */
-#ifdef DEBUG
- debug = (env != NULL && atoi(env) == 0) ? 0 : 1;
-#else
- debug = (env != NULL) ? 1 : 0;
-#endif
- }
-
- /* Now only print the string if we're required to do so. */
- if (debug) {
- fprintf(stderr, "%s: %s", prefixString, outputString);
- if (newline)
- fprintf(stderr, "\n");
-
-#if defined(_WIN32) && !defined(_WIN32_WCE)
- /* stderr from windows applications without console is not usually
- * visible, so communicate with the debugger instead */
- {
- char buf[4096];
- _mesa_snprintf(buf, sizeof(buf), "%s: %s%s", prefixString, outputString, newline ? "\n" : "");
- OutputDebugStringA(buf);
- }
-#endif
- }
-}
-
-
-/**
- * Return string version of GL error code.
- */
-static const char *
-error_string( GLenum error )
-{
- switch (error) {
- case GL_NO_ERROR:
- return "GL_NO_ERROR";
- case GL_INVALID_VALUE:
- return "GL_INVALID_VALUE";
- case GL_INVALID_ENUM:
- return "GL_INVALID_ENUM";
- case GL_INVALID_OPERATION:
- return "GL_INVALID_OPERATION";
- case GL_STACK_OVERFLOW:
- return "GL_STACK_OVERFLOW";
- case GL_STACK_UNDERFLOW:
- return "GL_STACK_UNDERFLOW";
- case GL_OUT_OF_MEMORY:
- return "GL_OUT_OF_MEMORY";
- case GL_TABLE_TOO_LARGE:
- return "GL_TABLE_TOO_LARGE";
- case GL_INVALID_FRAMEBUFFER_OPERATION_EXT:
- return "GL_INVALID_FRAMEBUFFER_OPERATION";
- default:
- return "unknown";
- }
-}
-
-
-/**
- * When a new type of error is recorded, print a message describing
- * previous errors which were accumulated.
- */
-static void
-flush_delayed_errors( struct gl_context *ctx )
-{
- char s[MAXSTRING];
-
- if (ctx->ErrorDebugCount) {
- _mesa_snprintf(s, MAXSTRING, "%d similar %s errors",
- ctx->ErrorDebugCount,
- error_string(ctx->ErrorValue));
-
- output_if_debug("Mesa", s, GL_TRUE);
-
- ctx->ErrorDebugCount = 0;
- }
-}
-
-
-/**
- * Report a warning (a recoverable error condition) to stderr if
- * either DEBUG is defined or the MESA_DEBUG env var is set.
- *
- * \param ctx GL context.
- * \param fmtString printf()-like format string.
- */
-void
-_mesa_warning( struct gl_context *ctx, const char *fmtString, ... )
-{
- char str[MAXSTRING];
- va_list args;
- va_start( args, fmtString );
- (void) vsnprintf( str, MAXSTRING, fmtString, args );
- va_end( args );
-
- if (ctx)
- flush_delayed_errors( ctx );
-
- output_if_debug("Mesa warning", str, GL_TRUE);
-}
-
-
-/**
- * Report an internal implementation problem.
- * Prints the message to stderr via fprintf().
- *
- * \param ctx GL context.
- * \param fmtString problem description string.
- */
-void
-_mesa_problem( const struct gl_context *ctx, const char *fmtString, ... )
-{
- va_list args;
- char str[MAXSTRING];
- (void) ctx;
-
- va_start( args, fmtString );
- vsnprintf( str, MAXSTRING, fmtString, args );
- va_end( args );
-
- fprintf(stderr, "Mesa %s implementation error: %s\n", MESA_VERSION_STRING, str);
- fprintf(stderr, "Please report at bugs.freedesktop.org\n");
-}
-
-
-/**
- * Record an OpenGL state error. These usually occur when the user
- * passes invalid parameters to a GL function.
- *
- * If debugging is enabled (either at compile-time via the DEBUG macro, or
- * run-time via the MESA_DEBUG environment variable), report the error with
- * _mesa_debug().
- *
- * \param ctx the GL context.
- * \param error the error value.
- * \param fmtString printf() style format string, followed by optional args
- */
-void
-_mesa_error( struct gl_context *ctx, GLenum error, const char *fmtString, ... )
-{
- static GLint debug = -1;
-
- /* Check debug environment variable only once:
- */
- if (debug == -1) {
- const char *debugEnv = _mesa_getenv("MESA_DEBUG");
-
-#ifdef DEBUG
- if (debugEnv && strstr(debugEnv, "silent"))
- debug = GL_FALSE;
- else
- debug = GL_TRUE;
-#else
- if (debugEnv)
- debug = GL_TRUE;
- else
- debug = GL_FALSE;
-#endif
- }
-
- if (debug) {
- if (ctx->ErrorValue == error &&
- ctx->ErrorDebugFmtString == fmtString) {
- ctx->ErrorDebugCount++;
- }
- else {
- char s[MAXSTRING], s2[MAXSTRING];
- va_list args;
-
- flush_delayed_errors( ctx );
-
- va_start(args, fmtString);
- vsnprintf(s, MAXSTRING, fmtString, args);
- va_end(args);
-
- _mesa_snprintf(s2, MAXSTRING, "%s in %s", error_string(error), s);
- output_if_debug("Mesa: User error", s2, GL_TRUE);
-
- ctx->ErrorDebugFmtString = fmtString;
- ctx->ErrorDebugCount = 0;
- }
- }
-
- _mesa_record_error(ctx, error);
-}
-
-
-/**
- * Report debug information. Print error message to stderr via fprintf().
- * No-op if DEBUG mode not enabled.
- *
- * \param ctx GL context.
- * \param fmtString printf()-style format string, followed by optional args.
- */
-void
-_mesa_debug( const struct gl_context *ctx, const char *fmtString, ... )
-{
-#ifdef DEBUG
- char s[MAXSTRING];
- va_list args;
- va_start(args, fmtString);
- vsnprintf(s, MAXSTRING, fmtString, args);
- va_end(args);
- output_if_debug("Mesa", s, GL_FALSE);
-#endif /* DEBUG */
- (void) ctx;
- (void) fmtString;
-}
-
-/*@}*/
+/** + * \file imports.c + * Standard C library function wrappers. + * + * Imports are services which the device driver or window system or + * operating system provides to the core renderer. The core renderer (Mesa) + * will call these functions in order to do memory allocation, simple I/O, + * etc. + * + * Some drivers will want to override/replace this file with something + * specialized, but that'll be rare. + * + * Eventually, I want to move roll the glheader.h file into this. + * + * \todo Functions still needed: + * - scanf + * - qsort + * - rand and RAND_MAX + */ + +/* + * Mesa 3-D graphics library + * Version: 7.1 + * + * Copyright (C) 1999-2007 Brian Paul All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included + * in all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN + * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN + * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + */ + + + +#include "imports.h" +#include "context.h" +#include "mtypes.h" +#include "version.h" + +#ifdef _GNU_SOURCE +#include <locale.h> +#ifdef __APPLE__ +#include <xlocale.h> +#endif +#endif + + +#define MAXSTRING 4000 /* for vsnprintf() */ + +#ifdef WIN32 +#define vsnprintf _vsnprintf +#elif defined(__IBMC__) || defined(__IBMCPP__) || ( defined(__VMS) && __CRTL_VER < 70312000 ) +extern int vsnprintf(char *str, size_t count, const char *fmt, va_list arg); +#ifdef __VMS +#include "vsnprintf.c" +#endif +#endif + +/**********************************************************************/ +/** \name Memory */ +/*@{*/ + +/** + * Allocate aligned memory. + * + * \param bytes number of bytes to allocate. + * \param alignment alignment (must be greater than zero). + * + * Allocates extra memory to accommodate rounding up the address for + * alignment and to record the real malloc address. + * + * \sa _mesa_align_free(). + */ +void * +_mesa_align_malloc(size_t bytes, unsigned long alignment) +{ +#if defined(HAVE_POSIX_MEMALIGN) + void *mem; + int err = posix_memalign(& mem, alignment, bytes); + if (err) + return NULL; + return mem; +#elif defined(_WIN32) && defined(_MSC_VER) + return _aligned_malloc(bytes, alignment); +#else + uintptr_t ptr, buf; + + ASSERT( alignment > 0 ); + + ptr = (uintptr_t) malloc(bytes + alignment + sizeof(void *)); + if (!ptr) + return NULL; + + buf = (ptr + alignment + sizeof(void *)) & ~(uintptr_t)(alignment - 1); + *(uintptr_t *)(buf - sizeof(void *)) = ptr; + +#ifdef DEBUG + /* mark the non-aligned area */ + while ( ptr < buf - sizeof(void *) ) { + *(unsigned long *)ptr = 0xcdcdcdcd; + ptr += sizeof(unsigned long); + } +#endif + + return (void *) buf; +#endif /* defined(HAVE_POSIX_MEMALIGN) */ +} + +/** + * Same as _mesa_align_malloc(), but using calloc(1, ) instead of + * malloc() + */ +void * +_mesa_align_calloc(size_t bytes, unsigned long alignment) +{ +#if defined(HAVE_POSIX_MEMALIGN) + void *mem; + + mem = _mesa_align_malloc(bytes, alignment); + if (mem != NULL) { + (void) memset(mem, 0, bytes); + } + + return mem; +#elif defined(_WIN32) && defined(_MSC_VER) + void *mem; + + mem = _aligned_malloc(bytes, alignment); + if (mem != NULL) { + (void) memset(mem, 0, bytes); + } + + return mem; +#else + uintptr_t ptr, buf; + + ASSERT( alignment > 0 ); + + ptr = (uintptr_t) calloc(1, bytes + alignment + sizeof(void *)); + if (!ptr) + return NULL; + + buf = (ptr + alignment + sizeof(void *)) & ~(uintptr_t)(alignment - 1); + *(uintptr_t *)(buf - sizeof(void *)) = ptr; + +#ifdef DEBUG + /* mark the non-aligned area */ + while ( ptr < buf - sizeof(void *) ) { + *(unsigned long *)ptr = 0xcdcdcdcd; + ptr += sizeof(unsigned long); + } +#endif + + return (void *)buf; +#endif /* defined(HAVE_POSIX_MEMALIGN) */ +} + +/** + * Free memory which was allocated with either _mesa_align_malloc() + * or _mesa_align_calloc(). + * \param ptr pointer to the memory to be freed. + * The actual address to free is stored in the word immediately before the + * address the client sees. + */ +void +_mesa_align_free(void *ptr) +{ +#if defined(HAVE_POSIX_MEMALIGN) + free(ptr); +#elif defined(_WIN32) && defined(_MSC_VER) + _aligned_free(ptr); +#else + void **cubbyHole = (void **) ((char *) ptr - sizeof(void *)); + void *realAddr = *cubbyHole; + free(realAddr); +#endif /* defined(HAVE_POSIX_MEMALIGN) */ +} + +/** + * Reallocate memory, with alignment. + */ +void * +_mesa_align_realloc(void *oldBuffer, size_t oldSize, size_t newSize, + unsigned long alignment) +{ +#if defined(_WIN32) && defined(_MSC_VER) + (void) oldSize; + return _aligned_realloc(oldBuffer, newSize, alignment); +#else + const size_t copySize = (oldSize < newSize) ? oldSize : newSize; + void *newBuf = _mesa_align_malloc(newSize, alignment); + if (newBuf && oldBuffer && copySize > 0) { + memcpy(newBuf, oldBuffer, copySize); + } + if (oldBuffer) + _mesa_align_free(oldBuffer); + return newBuf; +#endif +} + + + +/** Reallocate memory */ +void * +_mesa_realloc(void *oldBuffer, size_t oldSize, size_t newSize) +{ + const size_t copySize = (oldSize < newSize) ? oldSize : newSize; + void *newBuffer = malloc(newSize); + if (newBuffer && oldBuffer && copySize > 0) + memcpy(newBuffer, oldBuffer, copySize); + if (oldBuffer) + free(oldBuffer); + return newBuffer; +} + +/** + * Fill memory with a constant 16bit word. + * \param dst destination pointer. + * \param val value. + * \param n number of words. + */ +void +_mesa_memset16( unsigned short *dst, unsigned short val, size_t n ) +{ + while (n-- > 0) + *dst++ = val; +} + +/*@}*/ + + +/**********************************************************************/ +/** \name Math */ +/*@{*/ + +/** Wrapper around sqrt() */ +double +_mesa_sqrtd(double x) +{ + return sqrt(x); +} + + +/* + * A High Speed, Low Precision Square Root + * by Paul Lalonde and Robert Dawson + * from "Graphics Gems", Academic Press, 1990 + * + * SPARC implementation of a fast square root by table + * lookup. + * SPARC floating point format is as follows: + * + * BIT 31 30 23 22 0 + * sign exponent mantissa + */ +static short sqrttab[0x100]; /* declare table of square roots */ + +void +_mesa_init_sqrt_table(void) +{ +#if defined(USE_IEEE) && !defined(DEBUG) + unsigned short i; + fi_type fi; /* to access the bits of a float in C quickly */ + /* we use a union defined in glheader.h */ + + for(i=0; i<= 0x7f; i++) { + fi.i = 0; + + /* + * Build a float with the bit pattern i as mantissa + * and an exponent of 0, stored as 127 + */ + + fi.i = (i << 16) | (127 << 23); + fi.f = _mesa_sqrtd(fi.f); + + /* + * Take the square root then strip the first 7 bits of + * the mantissa into the table + */ + + sqrttab[i] = (fi.i & 0x7fffff) >> 16; + + /* + * Repeat the process, this time with an exponent of + * 1, stored as 128 + */ + + fi.i = 0; + fi.i = (i << 16) | (128 << 23); + fi.f = sqrt(fi.f); + sqrttab[i+0x80] = (fi.i & 0x7fffff) >> 16; + } +#else + (void) sqrttab; /* silence compiler warnings */ +#endif /*HAVE_FAST_MATH*/ +} + + +/** + * Single precision square root. + */ +float +_mesa_sqrtf( float x ) +{ +#if defined(USE_IEEE) && !defined(DEBUG) + fi_type num; + /* to access the bits of a float in C + * we use a union from glheader.h */ + + short e; /* the exponent */ + if (x == 0.0F) return 0.0F; /* check for square root of 0 */ + num.f = x; + e = (num.i >> 23) - 127; /* get the exponent - on a SPARC the */ + /* exponent is stored with 127 added */ + num.i &= 0x7fffff; /* leave only the mantissa */ + if (e & 0x01) num.i |= 0x800000; + /* the exponent is odd so we have to */ + /* look it up in the second half of */ + /* the lookup table, so we set the */ + /* high bit */ + e >>= 1; /* divide the exponent by two */ + /* note that in C the shift */ + /* operators are sign preserving */ + /* for signed operands */ + /* Do the table lookup, based on the quaternary mantissa, + * then reconstruct the result back into a float + */ + num.i = ((sqrttab[num.i >> 16]) << 16) | ((e + 127) << 23); + + return num.f; +#else + return (float) _mesa_sqrtd((double) x); +#endif +} + + +/** + inv_sqrt - A single precision 1/sqrt routine for IEEE format floats. + written by Josh Vanderhoof, based on newsgroup posts by James Van Buskirk + and Vesa Karvonen. +*/ +float +_mesa_inv_sqrtf(float n) +{ +#if defined(USE_IEEE) && !defined(DEBUG) + float r0, x0, y0; + float r1, x1, y1; + float r2, x2, y2; +#if 0 /* not used, see below -BP */ + float r3, x3, y3; +#endif + fi_type u; + unsigned int magic; + + /* + Exponent part of the magic number - + + We want to: + 1. subtract the bias from the exponent, + 2. negate it + 3. divide by two (rounding towards -inf) + 4. add the bias back + + Which is the same as subtracting the exponent from 381 and dividing + by 2. + + floor(-(x - 127) / 2) + 127 = floor((381 - x) / 2) + */ + + magic = 381 << 23; + + /* + Significand part of magic number - + + With the current magic number, "(magic - u.i) >> 1" will give you: + + for 1 <= u.f <= 2: 1.25 - u.f / 4 + for 2 <= u.f <= 4: 1.00 - u.f / 8 + + This isn't a bad approximation of 1/sqrt. The maximum difference from + 1/sqrt will be around .06. After three Newton-Raphson iterations, the + maximum difference is less than 4.5e-8. (Which is actually close + enough to make the following bias academic...) + + To get a better approximation you can add a bias to the magic + number. For example, if you subtract 1/2 of the maximum difference in + the first approximation (.03), you will get the following function: + + for 1 <= u.f <= 2: 1.22 - u.f / 4 + for 2 <= u.f <= 3.76: 0.97 - u.f / 8 + for 3.76 <= u.f <= 4: 0.72 - u.f / 16 + (The 3.76 to 4 range is where the result is < .5.) + + This is the closest possible initial approximation, but with a maximum + error of 8e-11 after three NR iterations, it is still not perfect. If + you subtract 0.0332281 instead of .03, the maximum error will be + 2.5e-11 after three NR iterations, which should be about as close as + is possible. + + for 1 <= u.f <= 2: 1.2167719 - u.f / 4 + for 2 <= u.f <= 3.73: 0.9667719 - u.f / 8 + for 3.73 <= u.f <= 4: 0.7167719 - u.f / 16 + + */ + + magic -= (int)(0.0332281 * (1 << 25)); + + u.f = n; + u.i = (magic - u.i) >> 1; + + /* + Instead of Newton-Raphson, we use Goldschmidt's algorithm, which + allows more parallelism. From what I understand, the parallelism + comes at the cost of less precision, because it lets error + accumulate across iterations. + */ + x0 = 1.0f; + y0 = 0.5f * n; + r0 = u.f; + + x1 = x0 * r0; + y1 = y0 * r0 * r0; + r1 = 1.5f - y1; + + x2 = x1 * r1; + y2 = y1 * r1 * r1; + r2 = 1.5f - y2; + +#if 1 + return x2 * r2; /* we can stop here, and be conformant -BP */ +#else + x3 = x2 * r2; + y3 = y2 * r2 * r2; + r3 = 1.5f - y3; + + return x3 * r3; +#endif +#else + return (float) (1.0 / sqrt(n)); +#endif +} + +#ifndef __GNUC__ +/** + * Find the first bit set in a word. + */ +int +_mesa_ffs(int32_t i) +{ +#if (defined(_WIN32) ) || defined(__IBMC__) || defined(__IBMCPP__) + register int bit = 0; + if (i != 0) { + if ((i & 0xffff) == 0) { + bit += 16; + i >>= 16; + } + if ((i & 0xff) == 0) { + bit += 8; + i >>= 8; + } + if ((i & 0xf) == 0) { + bit += 4; + i >>= 4; + } + while ((i & 1) == 0) { + bit++; + i >>= 1; + } + bit++; + } + return bit; +#else + return ffs(i); +#endif +} + + +/** + * Find position of first bit set in given value. + * XXX Warning: this function can only be used on 64-bit systems! + * \return position of least-significant bit set, starting at 1, return zero + * if no bits set. + */ +int +_mesa_ffsll(int64_t val) +{ + int bit; + + assert(sizeof(val) == 8); + + bit = _mesa_ffs((int32_t)val); + if (bit != 0) + return bit; + + bit = _mesa_ffs((int32_t)(val >> 32)); + if (bit != 0) + return 32 + bit; + + return 0; +} + + +#if ((_GNUC__ == 3 && __GNUC_MINOR__ < 4) || __GNUC__ < 4) +/** + * Return number of bits set in given GLuint. + */ +unsigned int +_mesa_bitcount(unsigned int n) +{ + unsigned int bits; + for (bits = 0; n > 0; n = n >> 1) { + bits += (n & 1); + } + return bits; +} +#endif +#endif + + +/** + * Convert a 4-byte float to a 2-byte half float. + * Based on code from: + * http://www.opengl.org/discussion_boards/ubb/Forum3/HTML/008786.html + */ +GLhalfARB +_mesa_float_to_half(float val) +{ + const fi_type fi = {val}; + const int flt_m = fi.i & 0x7fffff; + const int flt_e = (fi.i >> 23) & 0xff; + const int flt_s = (fi.i >> 31) & 0x1; + int s, e, m = 0; + GLhalfARB result; + + /* sign bit */ + s = flt_s; + + /* handle special cases */ + if ((flt_e == 0) && (flt_m == 0)) { + /* zero */ + /* m = 0; - already set */ + e = 0; + } + else if ((flt_e == 0) && (flt_m != 0)) { + /* denorm -- denorm float maps to 0 half */ + /* m = 0; - already set */ + e = 0; + } + else if ((flt_e == 0xff) && (flt_m == 0)) { + /* infinity */ + /* m = 0; - already set */ + e = 31; + } + else if ((flt_e == 0xff) && (flt_m != 0)) { + /* NaN */ + m = 1; + e = 31; + } + else { + /* regular number */ + const int new_exp = flt_e - 127; + if (new_exp < -24) { + /* this maps to 0 */ + /* m = 0; - already set */ + e = 0; + } + else if (new_exp < -14) { + /* this maps to a denorm */ + unsigned int exp_val = (unsigned int) (-14 - new_exp); /* 2^-exp_val*/ + e = 0; + switch (exp_val) { + case 0: + _mesa_warning(NULL, + "float_to_half: logical error in denorm creation!\n"); + /* m = 0; - already set */ + break; + case 1: m = 512 + (flt_m >> 14); break; + case 2: m = 256 + (flt_m >> 15); break; + case 3: m = 128 + (flt_m >> 16); break; + case 4: m = 64 + (flt_m >> 17); break; + case 5: m = 32 + (flt_m >> 18); break; + case 6: m = 16 + (flt_m >> 19); break; + case 7: m = 8 + (flt_m >> 20); break; + case 8: m = 4 + (flt_m >> 21); break; + case 9: m = 2 + (flt_m >> 22); break; + case 10: m = 1; break; + } + } + else if (new_exp > 15) { + /* map this value to infinity */ + /* m = 0; - already set */ + e = 31; + } + else { + /* regular */ + e = new_exp + 15; + m = flt_m >> 13; + } + } + + result = (s << 15) | (e << 10) | m; + return result; +} + + +/** + * Convert a 2-byte half float to a 4-byte float. + * Based on code from: + * http://www.opengl.org/discussion_boards/ubb/Forum3/HTML/008786.html + */ +float +_mesa_half_to_float(GLhalfARB val) +{ + /* XXX could also use a 64K-entry lookup table */ + const int m = val & 0x3ff; + const int e = (val >> 10) & 0x1f; + const int s = (val >> 15) & 0x1; + int flt_m, flt_e, flt_s; + fi_type fi; + float result; + + /* sign bit */ + flt_s = s; + + /* handle special cases */ + if ((e == 0) && (m == 0)) { + /* zero */ + flt_m = 0; + flt_e = 0; + } + else if ((e == 0) && (m != 0)) { + /* denorm -- denorm half will fit in non-denorm single */ + const float half_denorm = 1.0f / 16384.0f; /* 2^-14 */ + float mantissa = ((float) (m)) / 1024.0f; + float sign = s ? -1.0f : 1.0f; + return sign * mantissa * half_denorm; + } + else if ((e == 31) && (m == 0)) { + /* infinity */ + flt_e = 0xff; + flt_m = 0; + } + else if ((e == 31) && (m != 0)) { + /* NaN */ + flt_e = 0xff; + flt_m = 1; + } + else { + /* regular */ + flt_e = e + 112; + flt_m = m << 13; + } + + fi.i = (flt_s << 31) | (flt_e << 23) | flt_m; + result = fi.f; + return result; +} + +/*@}*/ + + +/**********************************************************************/ +/** \name Sort & Search */ +/*@{*/ + +/** + * Wrapper for bsearch(). + */ +void * +_mesa_bsearch( const void *key, const void *base, size_t nmemb, size_t size, + int (*compar)(const void *, const void *) ) +{ +#if defined(_WIN32_WCE) + void *mid; + int cmp; + while (nmemb) { + nmemb >>= 1; + mid = (char *)base + nmemb * size; + cmp = (*compar)(key, mid); + if (cmp == 0) + return mid; + if (cmp > 0) { + base = (char *)mid + size; + --nmemb; + } + } + return NULL; +#else + return bsearch(key, base, nmemb, size, compar); +#endif +} + +/*@}*/ + + +/**********************************************************************/ +/** \name Environment vars */ +/*@{*/ + +/** + * Wrapper for getenv(). + */ +char * +_mesa_getenv( const char *var ) +{ +#if defined(_XBOX) || defined(_WIN32_WCE) + return NULL; +#else + return getenv(var); +#endif +} + +/*@}*/ + + +/**********************************************************************/ +/** \name String */ +/*@{*/ + +/** + * Implemented using malloc() and strcpy. + * Note that NULL is handled accordingly. + */ +char * +_mesa_strdup( const char *s ) +{ + if (s) { + size_t l = strlen(s); + char *s2 = (char *) malloc(l + 1); + if (s2) + strcpy(s2, s); + return s2; + } + else { + return NULL; + } +} + +/** Wrapper around strtof() */ +float +_mesa_strtof( const char *s, char **end ) +{ +#if defined(_GNU_SOURCE) && !defined(__CYGWIN__) && !defined(__FreeBSD__) + static locale_t loc = NULL; + if (!loc) { + loc = newlocale(LC_CTYPE_MASK, "C", NULL); + } + return strtof_l(s, end, loc); +#elif defined(_ISOC99_SOURCE) || (defined(_XOPEN_SOURCE) && _XOPEN_SOURCE >= 600) + return strtof(s, end); +#else + return (float)strtod(s, end); +#endif +} + +/** Compute simple checksum/hash for a string */ +unsigned int +_mesa_str_checksum(const char *str) +{ + /* This could probably be much better */ + unsigned int sum, i; + const char *c; + sum = i = 1; + for (c = str; *c; c++, i++) + sum += *c * (i % 100); + return sum + i; +} + + +/*@}*/ + + +/** Wrapper around vsnprintf() */ +int +_mesa_snprintf( char *str, size_t size, const char *fmt, ... ) +{ + int r; + va_list args; + va_start( args, fmt ); + r = vsnprintf( str, size, fmt, args ); + va_end( args ); + return r; +} + + +/**********************************************************************/ +/** \name Diagnostics */ +/*@{*/ + +static void +output_if_debug(const char *prefixString, const char *outputString, + GLboolean newline) +{ + static int debug = -1; + + /* Check the MESA_DEBUG environment variable if it hasn't + * been checked yet. We only have to check it once... + */ + if (debug == -1) { + char *env = _mesa_getenv("MESA_DEBUG"); + + /* In a debug build, we print warning messages *unless* + * MESA_DEBUG is 0. In a non-debug build, we don't + * print warning messages *unless* MESA_DEBUG is + * set *to any value*. + */ +#ifdef DEBUG + debug = (env != NULL && atoi(env) == 0) ? 0 : 1; +#else + debug = (env != NULL) ? 1 : 0; +#endif + } + + /* Now only print the string if we're required to do so. */ + if (debug) { + fprintf(stderr, "%s: %s", prefixString, outputString); + if (newline) + fprintf(stderr, "\n"); + +#if defined(_WIN32) && !defined(_WIN32_WCE) + /* stderr from windows applications without console is not usually + * visible, so communicate with the debugger instead */ + { + char buf[4096]; + _mesa_snprintf(buf, sizeof(buf), "%s: %s%s", prefixString, outputString, newline ? "\n" : ""); + OutputDebugStringA(buf); + } +#endif + } +} + + +/** + * Return string version of GL error code. + */ +static const char * +error_string( GLenum error ) +{ + switch (error) { + case GL_NO_ERROR: + return "GL_NO_ERROR"; + case GL_INVALID_VALUE: + return "GL_INVALID_VALUE"; + case GL_INVALID_ENUM: + return "GL_INVALID_ENUM"; + case GL_INVALID_OPERATION: + return "GL_INVALID_OPERATION"; + case GL_STACK_OVERFLOW: + return "GL_STACK_OVERFLOW"; + case GL_STACK_UNDERFLOW: + return "GL_STACK_UNDERFLOW"; + case GL_OUT_OF_MEMORY: + return "GL_OUT_OF_MEMORY"; + case GL_TABLE_TOO_LARGE: + return "GL_TABLE_TOO_LARGE"; + case GL_INVALID_FRAMEBUFFER_OPERATION_EXT: + return "GL_INVALID_FRAMEBUFFER_OPERATION"; + default: + return "unknown"; + } +} + + +/** + * When a new type of error is recorded, print a message describing + * previous errors which were accumulated. + */ +static void +flush_delayed_errors( struct gl_context *ctx ) +{ + char s[MAXSTRING]; + + if (ctx->ErrorDebugCount) { + _mesa_snprintf(s, MAXSTRING, "%d similar %s errors", + ctx->ErrorDebugCount, + error_string(ctx->ErrorValue)); + + output_if_debug("Mesa", s, GL_TRUE); + + ctx->ErrorDebugCount = 0; + } +} + + +/** + * Report a warning (a recoverable error condition) to stderr if + * either DEBUG is defined or the MESA_DEBUG env var is set. + * + * \param ctx GL context. + * \param fmtString printf()-like format string. + */ +void +_mesa_warning( struct gl_context *ctx, const char *fmtString, ... ) +{ + char str[MAXSTRING]; + va_list args; + va_start( args, fmtString ); + (void) vsnprintf( str, MAXSTRING, fmtString, args ); + va_end( args ); + + if (ctx) + flush_delayed_errors( ctx ); + + output_if_debug("Mesa warning", str, GL_TRUE); +} + + +/** + * Report an internal implementation problem. + * Prints the message to stderr via fprintf(). + * + * \param ctx GL context. + * \param fmtString problem description string. + */ +void +_mesa_problem( const struct gl_context *ctx, const char *fmtString, ... ) +{ + va_list args; + char str[MAXSTRING]; + (void) ctx; + + va_start( args, fmtString ); + vsnprintf( str, MAXSTRING, fmtString, args ); + va_end( args ); + + fprintf(stderr, "Mesa %s implementation error: %s\n", MESA_VERSION_STRING, str); + fprintf(stderr, "Please report at bugs.freedesktop.org\n"); +} + + +/** + * Record an OpenGL state error. These usually occur when the user + * passes invalid parameters to a GL function. + * + * If debugging is enabled (either at compile-time via the DEBUG macro, or + * run-time via the MESA_DEBUG environment variable), report the error with + * _mesa_debug(). + * + * \param ctx the GL context. + * \param error the error value. + * \param fmtString printf() style format string, followed by optional args + */ +void +_mesa_error( struct gl_context *ctx, GLenum error, const char *fmtString, ... ) +{ + static GLint debug = -1; + + /* Check debug environment variable only once: + */ + if (debug == -1) { + const char *debugEnv = _mesa_getenv("MESA_DEBUG"); + +#ifdef DEBUG + if (debugEnv && strstr(debugEnv, "silent")) + debug = GL_FALSE; + else + debug = GL_TRUE; +#else + if (debugEnv) + debug = GL_TRUE; + else + debug = GL_FALSE; +#endif + } + + if (debug) { + if (ctx->ErrorValue == error && + ctx->ErrorDebugFmtString == fmtString) { + ctx->ErrorDebugCount++; + } + else { + char s[MAXSTRING], s2[MAXSTRING]; + va_list args; + + flush_delayed_errors( ctx ); + + va_start(args, fmtString); + vsnprintf(s, MAXSTRING, fmtString, args); + va_end(args); + + _mesa_snprintf(s2, MAXSTRING, "%s in %s", error_string(error), s); + output_if_debug("Mesa: User error", s2, GL_TRUE); + + ctx->ErrorDebugFmtString = fmtString; + ctx->ErrorDebugCount = 0; + } + } + + _mesa_record_error(ctx, error); +} + + +/** + * Report debug information. Print error message to stderr via fprintf(). + * No-op if DEBUG mode not enabled. + * + * \param ctx GL context. + * \param fmtString printf()-style format string, followed by optional args. + */ +void +_mesa_debug( const struct gl_context *ctx, const char *fmtString, ... ) +{ +#ifdef DEBUG + char s[MAXSTRING]; + va_list args; + va_start(args, fmtString); + vsnprintf(s, MAXSTRING, fmtString, args); + va_end(args); + output_if_debug("Mesa", s, GL_FALSE); +#endif /* DEBUG */ + (void) ctx; + (void) fmtString; +} + +/*@}*/ diff --git a/mesalib/src/mesa/main/imports.h b/mesalib/src/mesa/main/imports.h index af7a8cc00..a994dbcae 100644 --- a/mesalib/src/mesa/main/imports.h +++ b/mesalib/src/mesa/main/imports.h @@ -1,602 +1,615 @@ -/*
- * Mesa 3-D graphics library
- * Version: 7.5
- *
- * Copyright (C) 1999-2008 Brian Paul All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included
- * in all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
- * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
- * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
- * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- */
-
-
-/**
- * \file imports.h
- * Standard C library function wrappers.
- *
- * This file provides wrappers for all the standard C library functions
- * like malloc(), free(), printf(), getenv(), etc.
- */
-
-
-#ifndef IMPORTS_H
-#define IMPORTS_H
-
-
-#include "compiler.h"
-#include "glheader.h"
-
-
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-
-/**********************************************************************/
-/** Memory macros */
-/*@{*/
-
-/** Allocate \p BYTES bytes */
-#define MALLOC(BYTES) malloc(BYTES)
-/** Allocate and zero \p BYTES bytes */
-#define CALLOC(BYTES) calloc(1, BYTES)
-/** Allocate a structure of type \p T */
-#define MALLOC_STRUCT(T) (struct T *) malloc(sizeof(struct T))
-/** Allocate and zero a structure of type \p T */
-#define CALLOC_STRUCT(T) (struct T *) calloc(1, sizeof(struct T))
-/** Free memory */
-#define FREE(PTR) free(PTR)
-
-/*@}*/
-
-
-/*
- * For GL_ARB_vertex_buffer_object we need to treat vertex array pointers
- * as offsets into buffer stores. Since the vertex array pointer and
- * buffer store pointer are both pointers and we need to add them, we use
- * this macro.
- * Both pointers/offsets are expressed in bytes.
- */
-#define ADD_POINTERS(A, B) ( (GLubyte *) (A) + (uintptr_t) (B) )
-
-
-/**
- * Sometimes we treat GLfloats as GLints. On x86 systems, moving a float
- * as a int (thereby using integer registers instead of FP registers) is
- * a performance win. Typically, this can be done with ordinary casts.
- * But with gcc's -fstrict-aliasing flag (which defaults to on in gcc 3.0)
- * these casts generate warnings.
- * The following union typedef is used to solve that.
- */
-typedef union { GLfloat f; GLint i; } fi_type;
-
-
-
-/**********************************************************************
- * Math macros
- */
-
-#define MAX_GLUSHORT 0xffff
-#define MAX_GLUINT 0xffffffff
-
-/* Degrees to radians conversion: */
-#define DEG2RAD (M_PI/180.0)
-
-
-/***
- *** SQRTF: single-precision square root
- ***/
-#if 0 /* _mesa_sqrtf() not accurate enough - temporarily disabled */
-# define SQRTF(X) _mesa_sqrtf(X)
-#else
-# define SQRTF(X) (float) sqrt((float) (X))
-#endif
-
-
-/***
- *** INV_SQRTF: single-precision inverse square root
- ***/
-#if 0
-#define INV_SQRTF(X) _mesa_inv_sqrt(X)
-#else
-#define INV_SQRTF(X) (1.0F / SQRTF(X)) /* this is faster on a P4 */
-#endif
-
-
-/**
- * \name Work-arounds for platforms that lack C99 math functions
- */
-/*@{*/
-#if (!defined(_XOPEN_SOURCE) || (_XOPEN_SOURCE < 600)) && !defined(_ISOC99_SOURCE) \
- && (!defined(__STDC_VERSION__) || (__STDC_VERSION__ < 199901L)) \
- && (!defined(_MSC_VER) || (_MSC_VER < 1400))
-#define acosf(f) ((float) acos(f))
-#define asinf(f) ((float) asin(f))
-#define atan2f(x,y) ((float) atan2(x,y))
-#define atanf(f) ((float) atan(f))
-#define cielf(f) ((float) ciel(f))
-#define cosf(f) ((float) cos(f))
-#define coshf(f) ((float) cosh(f))
-#define expf(f) ((float) exp(f))
-#define exp2f(f) ((float) exp2(f))
-#define floorf(f) ((float) floor(f))
-#define logf(f) ((float) log(f))
-#define log2f(f) ((float) log2(f))
-#define powf(x,y) ((float) pow(x,y))
-#define sinf(f) ((float) sin(f))
-#define sinhf(f) ((float) sinh(f))
-#define sqrtf(f) ((float) sqrt(f))
-#define tanf(f) ((float) tan(f))
-#define tanhf(f) ((float) tanh(f))
-#define acoshf(f) ((float) acosh(f))
-#define asinhf(f) ((float) asinh(f))
-#define atanhf(f) ((float) atanh(f))
-#endif
-
-#if defined(_MSC_VER)
-static INLINE float truncf(float x) { return x < 0.0f ? ceilf(x) : floorf(x); }
-static INLINE float exp2f(float x) { return powf(2.0f, x); }
-static INLINE float log2f(float x) { return logf(x) * 1.442695041f; }
-static INLINE float asinhf(float x) { return logf(x + sqrtf(x * x + 1.0f)); }
-static INLINE float acoshf(float x) { return logf(x + sqrtf(x * x - 1.0f)); }
-static INLINE float atanhf(float x) { return (logf(1.0f + x) - logf(1.0f - x)) / 2.0f; }
-static INLINE int isblank(int ch) { return ch == ' ' || ch == '\t'; }
-#define strtoll(p, e, b) _strtoi64(p, e, b)
-#endif
-/*@}*/
-
-/***
- *** LOG2: Log base 2 of float
- ***/
-#ifdef USE_IEEE
-#if 0
-/* This is pretty fast, but not accurate enough (only 2 fractional bits).
- * Based on code from http://www.stereopsis.com/log2.html
- */
-static INLINE GLfloat LOG2(GLfloat x)
-{
- const GLfloat y = x * x * x * x;
- const GLuint ix = *((GLuint *) &y);
- const GLuint exp = (ix >> 23) & 0xFF;
- const GLint log2 = ((GLint) exp) - 127;
- return (GLfloat) log2 * (1.0 / 4.0); /* 4, because of x^4 above */
-}
-#endif
-/* Pretty fast, and accurate.
- * Based on code from http://www.flipcode.com/totd/
- */
-static INLINE GLfloat LOG2(GLfloat val)
-{
- fi_type num;
- GLint log_2;
- num.f = val;
- log_2 = ((num.i >> 23) & 255) - 128;
- num.i &= ~(255 << 23);
- num.i += 127 << 23;
- num.f = ((-1.0f/3) * num.f + 2) * num.f - 2.0f/3;
- return num.f + log_2;
-}
-#else
-/*
- * NOTE: log_base_2(x) = log(x) / log(2)
- * NOTE: 1.442695 = 1/log(2).
- */
-#define LOG2(x) ((GLfloat) (log(x) * 1.442695F))
-#endif
-
-
-/***
- *** IS_INF_OR_NAN: test if float is infinite or NaN
- ***/
-#ifdef USE_IEEE
-static INLINE int IS_INF_OR_NAN( float x )
-{
- fi_type tmp;
- tmp.f = x;
- return !(int)((unsigned int)((tmp.i & 0x7fffffff)-0x7f800000) >> 31);
-}
-#elif defined(isfinite)
-#define IS_INF_OR_NAN(x) (!isfinite(x))
-#elif defined(finite)
-#define IS_INF_OR_NAN(x) (!finite(x))
-#elif defined(__VMS)
-#define IS_INF_OR_NAN(x) (!finite(x))
-#elif defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L
-#define IS_INF_OR_NAN(x) (!isfinite(x))
-#else
-#define IS_INF_OR_NAN(x) (!finite(x))
-#endif
-
-
-/***
- *** IS_NEGATIVE: test if float is negative
- ***/
-#if defined(USE_IEEE)
-static INLINE int GET_FLOAT_BITS( float x )
-{
- fi_type fi;
- fi.f = x;
- return fi.i;
-}
-#define IS_NEGATIVE(x) (GET_FLOAT_BITS(x) < 0)
-#else
-#define IS_NEGATIVE(x) (x < 0.0F)
-#endif
-
-
-/***
- *** DIFFERENT_SIGNS: test if two floats have opposite signs
- ***/
-#if defined(USE_IEEE)
-#define DIFFERENT_SIGNS(x,y) ((GET_FLOAT_BITS(x) ^ GET_FLOAT_BITS(y)) & (1<<31))
-#else
-/* Could just use (x*y<0) except for the flatshading requirements.
- * Maybe there's a better way?
- */
-#define DIFFERENT_SIGNS(x,y) ((x) * (y) <= 0.0F && (x) - (y) != 0.0F)
-#endif
-
-
-/***
- *** CEILF: ceiling of float
- *** FLOORF: floor of float
- *** FABSF: absolute value of float
- *** LOGF: the natural logarithm (base e) of the value
- *** EXPF: raise e to the value
- *** LDEXPF: multiply value by an integral power of two
- *** FREXPF: extract mantissa and exponent from value
- ***/
-#if defined(__gnu_linux__)
-/* C99 functions */
-#define CEILF(x) ceilf(x)
-#define FLOORF(x) floorf(x)
-#define FABSF(x) fabsf(x)
-#define LOGF(x) logf(x)
-#define EXPF(x) expf(x)
-#define LDEXPF(x,y) ldexpf(x,y)
-#define FREXPF(x,y) frexpf(x,y)
-#else
-#define CEILF(x) ((GLfloat) ceil(x))
-#define FLOORF(x) ((GLfloat) floor(x))
-#define FABSF(x) ((GLfloat) fabs(x))
-#define LOGF(x) ((GLfloat) log(x))
-#define EXPF(x) ((GLfloat) exp(x))
-#define LDEXPF(x,y) ((GLfloat) ldexp(x,y))
-#define FREXPF(x,y) ((GLfloat) frexp(x,y))
-#endif
-
-
-/***
- *** IROUND: return (as an integer) float rounded to nearest integer
- ***/
-#if defined(USE_X86_ASM) && defined(__GNUC__) && defined(__i386__)
-static INLINE int iround(float f)
-{
- int r;
- __asm__ ("fistpl %0" : "=m" (r) : "t" (f) : "st");
- return r;
-}
-#define IROUND(x) iround(x)
-#elif defined(USE_X86_ASM) && defined(_MSC_VER)
-static INLINE int iround(float f)
-{
- int r;
- _asm {
- fld f
- fistp r
- }
- return r;
-}
-#define IROUND(x) iround(x)
-#elif defined(__WATCOMC__) && defined(__386__)
-long iround(float f);
-#pragma aux iround = \
- "push eax" \
- "fistp dword ptr [esp]" \
- "pop eax" \
- parm [8087] \
- value [eax] \
- modify exact [eax];
-#define IROUND(x) iround(x)
-#else
-#define IROUND(f) ((int) (((f) >= 0.0F) ? ((f) + 0.5F) : ((f) - 0.5F)))
-#endif
-
-#define IROUND64(f) ((GLint64) (((f) >= 0.0F) ? ((f) + 0.5F) : ((f) - 0.5F)))
-
-/***
- *** IROUND_POS: return (as an integer) positive float rounded to nearest int
- ***/
-#ifdef DEBUG
-#define IROUND_POS(f) (assert((f) >= 0.0F), IROUND(f))
-#else
-#define IROUND_POS(f) (IROUND(f))
-#endif
-
-
-/***
- *** IFLOOR: return (as an integer) floor of float
- ***/
-#if defined(USE_X86_ASM) && defined(__GNUC__) && defined(__i386__)
-/*
- * IEEE floor for computers that round to nearest or even.
- * 'f' must be between -4194304 and 4194303.
- * This floor operation is done by "(iround(f + .5) + iround(f - .5)) >> 1",
- * but uses some IEEE specific tricks for better speed.
- * Contributed by Josh Vanderhoof
- */
-static INLINE int ifloor(float f)
-{
- int ai, bi;
- double af, bf;
- af = (3 << 22) + 0.5 + (double)f;
- bf = (3 << 22) + 0.5 - (double)f;
- /* GCC generates an extra fstp/fld without this. */
- __asm__ ("fstps %0" : "=m" (ai) : "t" (af) : "st");
- __asm__ ("fstps %0" : "=m" (bi) : "t" (bf) : "st");
- return (ai - bi) >> 1;
-}
-#define IFLOOR(x) ifloor(x)
-#elif defined(USE_IEEE)
-static INLINE int ifloor(float f)
-{
- int ai, bi;
- double af, bf;
- fi_type u;
-
- af = (3 << 22) + 0.5 + (double)f;
- bf = (3 << 22) + 0.5 - (double)f;
- u.f = (float) af; ai = u.i;
- u.f = (float) bf; bi = u.i;
- return (ai - bi) >> 1;
-}
-#define IFLOOR(x) ifloor(x)
-#else
-static INLINE int ifloor(float f)
-{
- int i = IROUND(f);
- return (i > f) ? i - 1 : i;
-}
-#define IFLOOR(x) ifloor(x)
-#endif
-
-
-/***
- *** ICEIL: return (as an integer) ceiling of float
- ***/
-#if defined(USE_X86_ASM) && defined(__GNUC__) && defined(__i386__)
-/*
- * IEEE ceil for computers that round to nearest or even.
- * 'f' must be between -4194304 and 4194303.
- * This ceil operation is done by "(iround(f + .5) + iround(f - .5) + 1) >> 1",
- * but uses some IEEE specific tricks for better speed.
- * Contributed by Josh Vanderhoof
- */
-static INLINE int iceil(float f)
-{
- int ai, bi;
- double af, bf;
- af = (3 << 22) + 0.5 + (double)f;
- bf = (3 << 22) + 0.5 - (double)f;
- /* GCC generates an extra fstp/fld without this. */
- __asm__ ("fstps %0" : "=m" (ai) : "t" (af) : "st");
- __asm__ ("fstps %0" : "=m" (bi) : "t" (bf) : "st");
- return (ai - bi + 1) >> 1;
-}
-#define ICEIL(x) iceil(x)
-#elif defined(USE_IEEE)
-static INLINE int iceil(float f)
-{
- int ai, bi;
- double af, bf;
- fi_type u;
- af = (3 << 22) + 0.5 + (double)f;
- bf = (3 << 22) + 0.5 - (double)f;
- u.f = (float) af; ai = u.i;
- u.f = (float) bf; bi = u.i;
- return (ai - bi + 1) >> 1;
-}
-#define ICEIL(x) iceil(x)
-#else
-static INLINE int iceil(float f)
-{
- int i = IROUND(f);
- return (i < f) ? i + 1 : i;
-}
-#define ICEIL(x) iceil(x)
-#endif
-
-
-/**
- * Is x a power of two?
- */
-static INLINE int
-_mesa_is_pow_two(int x)
-{
- return !(x & (x - 1));
-}
-
-/**
- * Round given integer to next higer power of two
- * If X is zero result is undefined.
- *
- * Source for the fallback implementation is
- * Sean Eron Anderson's webpage "Bit Twiddling Hacks"
- * http://graphics.stanford.edu/~seander/bithacks.html
- *
- * When using builtin function have to do some work
- * for case when passed values 1 to prevent hiting
- * undefined result from __builtin_clz. Undefined
- * results would be different depending on optimization
- * level used for build.
- */
-static INLINE int32_t
-_mesa_next_pow_two_32(uint32_t x)
-{
-#if defined(__GNUC__) && \
- ((__GNUC__ == 3 && __GNUC_MINOR__ >= 4) || __GNUC__ >= 4)
- uint32_t y = (x != 1);
- return (1 + y) << ((__builtin_clz(x - y) ^ 31) );
-#else
- x--;
- x |= x >> 1;
- x |= x >> 2;
- x |= x >> 4;
- x |= x >> 8;
- x |= x >> 16;
- x++;
- return x;
-#endif
-}
-
-static INLINE int64_t
-_mesa_next_pow_two_64(uint64_t x)
-{
-#if defined(__GNUC__) && \
- ((__GNUC__ == 3 && __GNUC_MINOR__ >= 4) || __GNUC__ >= 4)
- uint64_t y = (x != 1);
- if (sizeof(x) == sizeof(long))
- return (1 + y) << ((__builtin_clzl(x - y) ^ 63));
- else
- return (1 + y) << ((__builtin_clzll(x - y) ^ 63));
-#else
- x--;
- x |= x >> 1;
- x |= x >> 2;
- x |= x >> 4;
- x |= x >> 8;
- x |= x >> 16;
- x |= x >> 32;
- x++;
- return x;
-#endif
-}
-
-
-/**
- * Return 1 if this is a little endian machine, 0 if big endian.
- */
-static INLINE GLboolean
-_mesa_little_endian(void)
-{
- const GLuint ui = 1; /* intentionally not static */
- return *((const GLubyte *) &ui);
-}
-
-
-
-/**********************************************************************
- * Functions
- */
-
-extern void *
-_mesa_align_malloc( size_t bytes, unsigned long alignment );
-
-extern void *
-_mesa_align_calloc( size_t bytes, unsigned long alignment );
-
-extern void
-_mesa_align_free( void *ptr );
-
-extern void *
-_mesa_align_realloc(void *oldBuffer, size_t oldSize, size_t newSize,
- unsigned long alignment);
-
-extern void *
-_mesa_exec_malloc( GLuint size );
-
-extern void
-_mesa_exec_free( void *addr );
-
-extern void *
-_mesa_realloc( void *oldBuffer, size_t oldSize, size_t newSize );
-
-extern void
-_mesa_memset16( unsigned short *dst, unsigned short val, size_t n );
-
-extern double
-_mesa_sqrtd(double x);
-
-extern float
-_mesa_sqrtf(float x);
-
-extern float
-_mesa_inv_sqrtf(float x);
-
-extern void
-_mesa_init_sqrt_table(void);
-
-extern int
-_mesa_ffs(int32_t i);
-
-extern int
-_mesa_ffsll(int64_t i);
-
-extern unsigned int
-_mesa_bitcount(unsigned int n);
-
-extern GLhalfARB
-_mesa_float_to_half(float f);
-
-extern float
-_mesa_half_to_float(GLhalfARB h);
-
-
-extern void *
-_mesa_bsearch( const void *key, const void *base, size_t nmemb, size_t size,
- int (*compar)(const void *, const void *) );
-
-extern char *
-_mesa_getenv( const char *var );
-
-extern char *
-_mesa_strdup( const char *s );
-
-extern float
-_mesa_strtof( const char *s, char **end );
-
-extern unsigned int
-_mesa_str_checksum(const char *str);
-
-extern int
-_mesa_snprintf( char *str, size_t size, const char *fmt, ... ) PRINTFLIKE(3, 4);
-
-struct gl_context;
-
-extern void
-_mesa_warning( struct gl_context *gc, const char *fmtString, ... ) PRINTFLIKE(2, 3);
-
-extern void
-_mesa_problem( const struct gl_context *ctx, const char *fmtString, ... ) PRINTFLIKE(2, 3);
-
-extern void
-_mesa_error( struct gl_context *ctx, GLenum error, const char *fmtString, ... ) PRINTFLIKE(3, 4);
-
-extern void
-_mesa_debug( const struct gl_context *ctx, const char *fmtString, ... ) PRINTFLIKE(2, 3);
-
-
-#if defined(_MSC_VER) && !defined(snprintf)
-#define snprintf _snprintf
-#endif
-
-
-#ifdef __cplusplus
-}
-#endif
-
-
-#endif /* IMPORTS_H */
+/* + * Mesa 3-D graphics library + * Version: 7.5 + * + * Copyright (C) 1999-2008 Brian Paul All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included + * in all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN + * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN + * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + */ + + +/** + * \file imports.h + * Standard C library function wrappers. + * + * This file provides wrappers for all the standard C library functions + * like malloc(), free(), printf(), getenv(), etc. + */ + + +#ifndef IMPORTS_H +#define IMPORTS_H + + +#include "compiler.h" +#include "glheader.h" + + +#ifdef __cplusplus +extern "C" { +#endif + + +/**********************************************************************/ +/** Memory macros */ +/*@{*/ + +/** Allocate \p BYTES bytes */ +#define MALLOC(BYTES) malloc(BYTES) +/** Allocate and zero \p BYTES bytes */ +#define CALLOC(BYTES) calloc(1, BYTES) +/** Allocate a structure of type \p T */ +#define MALLOC_STRUCT(T) (struct T *) malloc(sizeof(struct T)) +/** Allocate and zero a structure of type \p T */ +#define CALLOC_STRUCT(T) (struct T *) calloc(1, sizeof(struct T)) +/** Free memory */ +#define FREE(PTR) free(PTR) + +/*@}*/ + + +/* + * For GL_ARB_vertex_buffer_object we need to treat vertex array pointers + * as offsets into buffer stores. Since the vertex array pointer and + * buffer store pointer are both pointers and we need to add them, we use + * this macro. + * Both pointers/offsets are expressed in bytes. + */ +#define ADD_POINTERS(A, B) ( (GLubyte *) (A) + (uintptr_t) (B) ) + + +/** + * Sometimes we treat GLfloats as GLints. On x86 systems, moving a float + * as a int (thereby using integer registers instead of FP registers) is + * a performance win. Typically, this can be done with ordinary casts. + * But with gcc's -fstrict-aliasing flag (which defaults to on in gcc 3.0) + * these casts generate warnings. + * The following union typedef is used to solve that. + */ +typedef union { GLfloat f; GLint i; } fi_type; + + + +/********************************************************************** + * Math macros + */ + +#define MAX_GLUSHORT 0xffff +#define MAX_GLUINT 0xffffffff + +/* Degrees to radians conversion: */ +#define DEG2RAD (M_PI/180.0) + + +/*** + *** SQRTF: single-precision square root + ***/ +#if 0 /* _mesa_sqrtf() not accurate enough - temporarily disabled */ +# define SQRTF(X) _mesa_sqrtf(X) +#else +# define SQRTF(X) (float) sqrt((float) (X)) +#endif + + +/*** + *** INV_SQRTF: single-precision inverse square root + ***/ +#if 0 +#define INV_SQRTF(X) _mesa_inv_sqrt(X) +#else +#define INV_SQRTF(X) (1.0F / SQRTF(X)) /* this is faster on a P4 */ +#endif + + +/** + * \name Work-arounds for platforms that lack C99 math functions + */ +/*@{*/ +#if (!defined(_XOPEN_SOURCE) || (_XOPEN_SOURCE < 600)) && !defined(_ISOC99_SOURCE) \ + && (!defined(__STDC_VERSION__) || (__STDC_VERSION__ < 199901L)) \ + && (!defined(_MSC_VER) || (_MSC_VER < 1400)) +#define acosf(f) ((float) acos(f)) +#define asinf(f) ((float) asin(f)) +#define atan2f(x,y) ((float) atan2(x,y)) +#define atanf(f) ((float) atan(f)) +#define cielf(f) ((float) ciel(f)) +#define cosf(f) ((float) cos(f)) +#define coshf(f) ((float) cosh(f)) +#define expf(f) ((float) exp(f)) +#define exp2f(f) ((float) exp2(f)) +#define floorf(f) ((float) floor(f)) +#define logf(f) ((float) log(f)) +#define log2f(f) ((float) log2(f)) +#define powf(x,y) ((float) pow(x,y)) +#define sinf(f) ((float) sin(f)) +#define sinhf(f) ((float) sinh(f)) +#define sqrtf(f) ((float) sqrt(f)) +#define tanf(f) ((float) tan(f)) +#define tanhf(f) ((float) tanh(f)) +#define acoshf(f) ((float) acosh(f)) +#define asinhf(f) ((float) asinh(f)) +#define atanhf(f) ((float) atanh(f)) +#endif + +#if defined(_MSC_VER) +static INLINE float truncf(float x) { return x < 0.0f ? ceilf(x) : floorf(x); } +static INLINE float exp2f(float x) { return powf(2.0f, x); } +static INLINE float log2f(float x) { return logf(x) * 1.442695041f; } +static INLINE float asinhf(float x) { return logf(x + sqrtf(x * x + 1.0f)); } +static INLINE float acoshf(float x) { return logf(x + sqrtf(x * x - 1.0f)); } +static INLINE float atanhf(float x) { return (logf(1.0f + x) - logf(1.0f - x)) / 2.0f; } +static INLINE int isblank(int ch) { return ch == ' ' || ch == '\t'; } +#define strtoll(p, e, b) _strtoi64(p, e, b) +#endif +/*@}*/ + +/*** + *** LOG2: Log base 2 of float + ***/ +#ifdef USE_IEEE +#if 0 +/* This is pretty fast, but not accurate enough (only 2 fractional bits). + * Based on code from http://www.stereopsis.com/log2.html + */ +static INLINE GLfloat LOG2(GLfloat x) +{ + const GLfloat y = x * x * x * x; + const GLuint ix = *((GLuint *) &y); + const GLuint exp = (ix >> 23) & 0xFF; + const GLint log2 = ((GLint) exp) - 127; + return (GLfloat) log2 * (1.0 / 4.0); /* 4, because of x^4 above */ +} +#endif +/* Pretty fast, and accurate. + * Based on code from http://www.flipcode.com/totd/ + */ +static INLINE GLfloat LOG2(GLfloat val) +{ + fi_type num; + GLint log_2; + num.f = val; + log_2 = ((num.i >> 23) & 255) - 128; + num.i &= ~(255 << 23); + num.i += 127 << 23; + num.f = ((-1.0f/3) * num.f + 2) * num.f - 2.0f/3; + return num.f + log_2; +} +#else +/* + * NOTE: log_base_2(x) = log(x) / log(2) + * NOTE: 1.442695 = 1/log(2). + */ +#define LOG2(x) ((GLfloat) (log(x) * 1.442695F)) +#endif + + +/*** + *** IS_INF_OR_NAN: test if float is infinite or NaN + ***/ +#ifdef USE_IEEE +static INLINE int IS_INF_OR_NAN( float x ) +{ + fi_type tmp; + tmp.f = x; + return !(int)((unsigned int)((tmp.i & 0x7fffffff)-0x7f800000) >> 31); +} +#elif defined(isfinite) +#define IS_INF_OR_NAN(x) (!isfinite(x)) +#elif defined(finite) +#define IS_INF_OR_NAN(x) (!finite(x)) +#elif defined(__VMS) +#define IS_INF_OR_NAN(x) (!finite(x)) +#elif defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L +#define IS_INF_OR_NAN(x) (!isfinite(x)) +#else +#define IS_INF_OR_NAN(x) (!finite(x)) +#endif + + +/*** + *** IS_NEGATIVE: test if float is negative + ***/ +#if defined(USE_IEEE) +static INLINE int GET_FLOAT_BITS( float x ) +{ + fi_type fi; + fi.f = x; + return fi.i; +} +#define IS_NEGATIVE(x) (GET_FLOAT_BITS(x) < 0) +#else +#define IS_NEGATIVE(x) (x < 0.0F) +#endif + + +/*** + *** DIFFERENT_SIGNS: test if two floats have opposite signs + ***/ +#if defined(USE_IEEE) +#define DIFFERENT_SIGNS(x,y) ((GET_FLOAT_BITS(x) ^ GET_FLOAT_BITS(y)) & (1<<31)) +#else +/* Could just use (x*y<0) except for the flatshading requirements. + * Maybe there's a better way? + */ +#define DIFFERENT_SIGNS(x,y) ((x) * (y) <= 0.0F && (x) - (y) != 0.0F) +#endif + + +/*** + *** CEILF: ceiling of float + *** FLOORF: floor of float + *** FABSF: absolute value of float + *** LOGF: the natural logarithm (base e) of the value + *** EXPF: raise e to the value + *** LDEXPF: multiply value by an integral power of two + *** FREXPF: extract mantissa and exponent from value + ***/ +#if defined(__gnu_linux__) +/* C99 functions */ +#define CEILF(x) ceilf(x) +#define FLOORF(x) floorf(x) +#define FABSF(x) fabsf(x) +#define LOGF(x) logf(x) +#define EXPF(x) expf(x) +#define LDEXPF(x,y) ldexpf(x,y) +#define FREXPF(x,y) frexpf(x,y) +#else +#define CEILF(x) ((GLfloat) ceil(x)) +#define FLOORF(x) ((GLfloat) floor(x)) +#define FABSF(x) ((GLfloat) fabs(x)) +#define LOGF(x) ((GLfloat) log(x)) +#define EXPF(x) ((GLfloat) exp(x)) +#define LDEXPF(x,y) ((GLfloat) ldexp(x,y)) +#define FREXPF(x,y) ((GLfloat) frexp(x,y)) +#endif + + +/*** + *** IROUND: return (as an integer) float rounded to nearest integer + ***/ +#if defined(USE_X86_ASM) && defined(__GNUC__) && defined(__i386__) +static INLINE int iround(float f) +{ + int r; + __asm__ ("fistpl %0" : "=m" (r) : "t" (f) : "st"); + return r; +} +#define IROUND(x) iround(x) +#elif defined(USE_X86_ASM) && defined(_MSC_VER) +static INLINE int iround(float f) +{ + int r; + _asm { + fld f + fistp r + } + return r; +} +#define IROUND(x) iround(x) +#elif defined(__WATCOMC__) && defined(__386__) +long iround(float f); +#pragma aux iround = \ + "push eax" \ + "fistp dword ptr [esp]" \ + "pop eax" \ + parm [8087] \ + value [eax] \ + modify exact [eax]; +#define IROUND(x) iround(x) +#else +#define IROUND(f) ((int) (((f) >= 0.0F) ? ((f) + 0.5F) : ((f) - 0.5F))) +#endif + +#define IROUND64(f) ((GLint64) (((f) >= 0.0F) ? ((f) + 0.5F) : ((f) - 0.5F))) + +/*** + *** IROUND_POS: return (as an integer) positive float rounded to nearest int + ***/ +#ifdef DEBUG +#define IROUND_POS(f) (assert((f) >= 0.0F), IROUND(f)) +#else +#define IROUND_POS(f) (IROUND(f)) +#endif + + +/*** + *** IFLOOR: return (as an integer) floor of float + ***/ +#if defined(USE_X86_ASM) && defined(__GNUC__) && defined(__i386__) +/* + * IEEE floor for computers that round to nearest or even. + * 'f' must be between -4194304 and 4194303. + * This floor operation is done by "(iround(f + .5) + iround(f - .5)) >> 1", + * but uses some IEEE specific tricks for better speed. + * Contributed by Josh Vanderhoof + */ +static INLINE int ifloor(float f) +{ + int ai, bi; + double af, bf; + af = (3 << 22) + 0.5 + (double)f; + bf = (3 << 22) + 0.5 - (double)f; + /* GCC generates an extra fstp/fld without this. */ + __asm__ ("fstps %0" : "=m" (ai) : "t" (af) : "st"); + __asm__ ("fstps %0" : "=m" (bi) : "t" (bf) : "st"); + return (ai - bi) >> 1; +} +#define IFLOOR(x) ifloor(x) +#elif defined(USE_IEEE) +static INLINE int ifloor(float f) +{ + int ai, bi; + double af, bf; + fi_type u; + + af = (3 << 22) + 0.5 + (double)f; + bf = (3 << 22) + 0.5 - (double)f; + u.f = (float) af; ai = u.i; + u.f = (float) bf; bi = u.i; + return (ai - bi) >> 1; +} +#define IFLOOR(x) ifloor(x) +#else +static INLINE int ifloor(float f) +{ + int i = IROUND(f); + return (i > f) ? i - 1 : i; +} +#define IFLOOR(x) ifloor(x) +#endif + + +/*** + *** ICEIL: return (as an integer) ceiling of float + ***/ +#if defined(USE_X86_ASM) && defined(__GNUC__) && defined(__i386__) +/* + * IEEE ceil for computers that round to nearest or even. + * 'f' must be between -4194304 and 4194303. + * This ceil operation is done by "(iround(f + .5) + iround(f - .5) + 1) >> 1", + * but uses some IEEE specific tricks for better speed. + * Contributed by Josh Vanderhoof + */ +static INLINE int iceil(float f) +{ + int ai, bi; + double af, bf; + af = (3 << 22) + 0.5 + (double)f; + bf = (3 << 22) + 0.5 - (double)f; + /* GCC generates an extra fstp/fld without this. */ + __asm__ ("fstps %0" : "=m" (ai) : "t" (af) : "st"); + __asm__ ("fstps %0" : "=m" (bi) : "t" (bf) : "st"); + return (ai - bi + 1) >> 1; +} +#define ICEIL(x) iceil(x) +#elif defined(USE_IEEE) +static INLINE int iceil(float f) +{ + int ai, bi; + double af, bf; + fi_type u; + af = (3 << 22) + 0.5 + (double)f; + bf = (3 << 22) + 0.5 - (double)f; + u.f = (float) af; ai = u.i; + u.f = (float) bf; bi = u.i; + return (ai - bi + 1) >> 1; +} +#define ICEIL(x) iceil(x) +#else +static INLINE int iceil(float f) +{ + int i = IROUND(f); + return (i < f) ? i + 1 : i; +} +#define ICEIL(x) iceil(x) +#endif + + +/** + * Is x a power of two? + */ +static INLINE int +_mesa_is_pow_two(int x) +{ + return !(x & (x - 1)); +} + +/** + * Round given integer to next higer power of two + * If X is zero result is undefined. + * + * Source for the fallback implementation is + * Sean Eron Anderson's webpage "Bit Twiddling Hacks" + * http://graphics.stanford.edu/~seander/bithacks.html + * + * When using builtin function have to do some work + * for case when passed values 1 to prevent hiting + * undefined result from __builtin_clz. Undefined + * results would be different depending on optimization + * level used for build. + */ +static INLINE int32_t +_mesa_next_pow_two_32(uint32_t x) +{ +#if defined(__GNUC__) && \ + ((__GNUC__ == 3 && __GNUC_MINOR__ >= 4) || __GNUC__ >= 4) + uint32_t y = (x != 1); + return (1 + y) << ((__builtin_clz(x - y) ^ 31) ); +#else + x--; + x |= x >> 1; + x |= x >> 2; + x |= x >> 4; + x |= x >> 8; + x |= x >> 16; + x++; + return x; +#endif +} + +static INLINE int64_t +_mesa_next_pow_two_64(uint64_t x) +{ +#if defined(__GNUC__) && \ + ((__GNUC__ == 3 && __GNUC_MINOR__ >= 4) || __GNUC__ >= 4) + uint64_t y = (x != 1); + if (sizeof(x) == sizeof(long)) + return (1 + y) << ((__builtin_clzl(x - y) ^ 63)); + else + return (1 + y) << ((__builtin_clzll(x - y) ^ 63)); +#else + x--; + x |= x >> 1; + x |= x >> 2; + x |= x >> 4; + x |= x >> 8; + x |= x >> 16; + x |= x >> 32; + x++; + return x; +#endif +} + + +/** + * Return 1 if this is a little endian machine, 0 if big endian. + */ +static INLINE GLboolean +_mesa_little_endian(void) +{ + const GLuint ui = 1; /* intentionally not static */ + return *((const GLubyte *) &ui); +} + + + +/********************************************************************** + * Functions + */ + +extern void * +_mesa_align_malloc( size_t bytes, unsigned long alignment ); + +extern void * +_mesa_align_calloc( size_t bytes, unsigned long alignment ); + +extern void +_mesa_align_free( void *ptr ); + +extern void * +_mesa_align_realloc(void *oldBuffer, size_t oldSize, size_t newSize, + unsigned long alignment); + +extern void * +_mesa_exec_malloc( GLuint size ); + +extern void +_mesa_exec_free( void *addr ); + +extern void * +_mesa_realloc( void *oldBuffer, size_t oldSize, size_t newSize ); + +extern void +_mesa_memset16( unsigned short *dst, unsigned short val, size_t n ); + +extern double +_mesa_sqrtd(double x); + +extern float +_mesa_sqrtf(float x); + +extern float +_mesa_inv_sqrtf(float x); + +extern void +_mesa_init_sqrt_table(void); + +#ifdef __GNUC__ +#define _mesa_ffs(i) ffs(i) +#define _mesa_ffsll(i) ffsll(i) + +#if ((_GNUC__ == 3 && __GNUC_MINOR__ >= 4) || __GNUC__ >= 4) +#define _mesa_bitcount(i) __builtin_popcount(i) +#else +extern unsigned int +_mesa_bitcount(unsigned int n); +#endif + +#else +extern int +_mesa_ffs(int32_t i); + +extern int +_mesa_ffsll(int64_t i); + +extern unsigned int +_mesa_bitcount(unsigned int n); +#endif + +extern GLhalfARB +_mesa_float_to_half(float f); + +extern float +_mesa_half_to_float(GLhalfARB h); + + +extern void * +_mesa_bsearch( const void *key, const void *base, size_t nmemb, size_t size, + int (*compar)(const void *, const void *) ); + +extern char * +_mesa_getenv( const char *var ); + +extern char * +_mesa_strdup( const char *s ); + +extern float +_mesa_strtof( const char *s, char **end ); + +extern unsigned int +_mesa_str_checksum(const char *str); + +extern int +_mesa_snprintf( char *str, size_t size, const char *fmt, ... ) PRINTFLIKE(3, 4); + +struct gl_context; + +extern void +_mesa_warning( struct gl_context *gc, const char *fmtString, ... ) PRINTFLIKE(2, 3); + +extern void +_mesa_problem( const struct gl_context *ctx, const char *fmtString, ... ) PRINTFLIKE(2, 3); + +extern void +_mesa_error( struct gl_context *ctx, GLenum error, const char *fmtString, ... ) PRINTFLIKE(3, 4); + +extern void +_mesa_debug( const struct gl_context *ctx, const char *fmtString, ... ) PRINTFLIKE(2, 3); + + +#if defined(_MSC_VER) && !defined(snprintf) +#define snprintf _snprintf +#endif + + +#ifdef __cplusplus +} +#endif + + +#endif /* IMPORTS_H */ diff --git a/mesalib/src/mesa/program/sampler.cpp b/mesalib/src/mesa/program/sampler.cpp index 12c4a40a2..1457d1199 100644 --- a/mesalib/src/mesa/program/sampler.cpp +++ b/mesalib/src/mesa/program/sampler.cpp @@ -23,7 +23,6 @@ * DEALINGS IN THE SOFTWARE. */ -#include <cstdio> #include "ir.h" #include "glsl_types.h" #include "ir_visitor.h" diff --git a/mesalib/src/mesa/state_tracker/st_atom_sampler.c b/mesalib/src/mesa/state_tracker/st_atom_sampler.c index c9dfe248f..474cbd589 100644 --- a/mesalib/src/mesa/state_tracker/st_atom_sampler.c +++ b/mesalib/src/mesa/state_tracker/st_atom_sampler.c @@ -1,270 +1,228 @@ -/**************************************************************************
- *
- * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- **************************************************************************/
-
- /*
- * Authors:
- * Keith Whitwell <keith@tungstengraphics.com>
- * Brian Paul
- */
-
-
-#include "main/macros.h"
-
-#include "st_context.h"
-#include "st_cb_texture.h"
-#include "st_atom.h"
-#include "pipe/p_context.h"
-#include "pipe/p_defines.h"
-
-#include "cso_cache/cso_context.h"
-
-
-/**
- * Convert GLenum texcoord wrap tokens to pipe tokens.
- */
-static GLuint
-gl_wrap_xlate(GLenum wrap)
-{
- switch (wrap) {
- case GL_REPEAT:
- return PIPE_TEX_WRAP_REPEAT;
- case GL_CLAMP:
- return PIPE_TEX_WRAP_CLAMP;
- case GL_CLAMP_TO_EDGE:
- return PIPE_TEX_WRAP_CLAMP_TO_EDGE;
- case GL_CLAMP_TO_BORDER:
- return PIPE_TEX_WRAP_CLAMP_TO_BORDER;
- case GL_MIRRORED_REPEAT:
- return PIPE_TEX_WRAP_MIRROR_REPEAT;
- case GL_MIRROR_CLAMP_EXT:
- return PIPE_TEX_WRAP_MIRROR_CLAMP;
- case GL_MIRROR_CLAMP_TO_EDGE_EXT:
- return PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE;
- case GL_MIRROR_CLAMP_TO_BORDER_EXT:
- return PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER;
- default:
- assert(0);
- return 0;
- }
-}
-
-
-static GLuint
-gl_filter_to_mip_filter(GLenum filter)
-{
- switch (filter) {
- case GL_NEAREST:
- case GL_LINEAR:
- return PIPE_TEX_MIPFILTER_NONE;
-
- case GL_NEAREST_MIPMAP_NEAREST:
- case GL_LINEAR_MIPMAP_NEAREST:
- return PIPE_TEX_MIPFILTER_NEAREST;
-
- case GL_NEAREST_MIPMAP_LINEAR:
- case GL_LINEAR_MIPMAP_LINEAR:
- return PIPE_TEX_MIPFILTER_LINEAR;
-
- default:
- assert(0);
- return PIPE_TEX_MIPFILTER_NONE;
- }
-}
-
-
-static GLuint
-gl_filter_to_img_filter(GLenum filter)
-{
- switch (filter) {
- case GL_NEAREST:
- case GL_NEAREST_MIPMAP_NEAREST:
- case GL_NEAREST_MIPMAP_LINEAR:
- return PIPE_TEX_FILTER_NEAREST;
-
- case GL_LINEAR:
- case GL_LINEAR_MIPMAP_NEAREST:
- case GL_LINEAR_MIPMAP_LINEAR:
- return PIPE_TEX_FILTER_LINEAR;
-
- default:
- assert(0);
- return PIPE_TEX_FILTER_NEAREST;
- }
-}
-
-
-static void
-xlate_border_color(const GLfloat *colorIn, GLenum baseFormat, GLfloat *colorOut)
-{
- switch (baseFormat) {
- case GL_RED:
- colorOut[0] = colorIn[0];
- colorOut[1] = 0.0F;
- colorOut[2] = 0.0F;
- colorOut[3] = 1.0F;
- break;
- case GL_RG:
- colorOut[0] = colorIn[0];
- colorOut[1] = colorIn[1];
- colorOut[2] = 0.0F;
- colorOut[3] = 1.0F;
- break;
- case GL_RGB:
- colorOut[0] = colorIn[0];
- colorOut[1] = colorIn[1];
- colorOut[2] = colorIn[2];
- colorOut[3] = 1.0F;
- break;
- case GL_ALPHA:
- colorOut[0] = colorOut[1] = colorOut[2] = 0.0;
- colorOut[3] = colorIn[3];
- break;
- case GL_LUMINANCE:
- colorOut[0] = colorOut[1] = colorOut[2] = colorIn[0];
- colorOut[3] = 1.0;
- break;
- case GL_LUMINANCE_ALPHA:
- colorOut[0] = colorOut[1] = colorOut[2] = colorIn[0];
- colorOut[3] = colorIn[3];
- break;
- case GL_INTENSITY:
- colorOut[0] = colorOut[1] = colorOut[2] = colorOut[3] = colorIn[0];
- break;
- default:
- COPY_4V(colorOut, colorIn);
- }
-}
-
-
-static void
-update_samplers(struct st_context *st)
-{
- struct gl_vertex_program *vprog = st->ctx->VertexProgram._Current;
- struct gl_fragment_program *fprog = st->ctx->FragmentProgram._Current;
- const GLbitfield samplersUsed = (vprog->Base.SamplersUsed |
- fprog->Base.SamplersUsed);
- GLuint su;
-
- st->state.num_samplers = 0;
-
- /* loop over sampler units (aka tex image units) */
- for (su = 0; su < st->ctx->Const.MaxTextureImageUnits; su++) {
- struct pipe_sampler_state *sampler = st->state.samplers + su;
-
- memset(sampler, 0, sizeof(*sampler));
-
- if (samplersUsed & (1 << su)) {
- struct gl_texture_object *texobj;
- struct gl_texture_image *teximg;
- GLuint texUnit;
-
- if (fprog->Base.SamplersUsed & (1 << su))
- texUnit = fprog->Base.SamplerUnits[su];
- else
- texUnit = vprog->Base.SamplerUnits[su];
-
- texobj = st->ctx->Texture.Unit[texUnit]._Current;
- if (!texobj) {
- texobj = st_get_default_texture(st);
- }
-
- teximg = texobj->Image[0][texobj->BaseLevel];
-
- sampler->wrap_s = gl_wrap_xlate(texobj->WrapS);
- sampler->wrap_t = gl_wrap_xlate(texobj->WrapT);
- sampler->wrap_r = gl_wrap_xlate(texobj->WrapR);
-
- sampler->min_img_filter = gl_filter_to_img_filter(texobj->MinFilter);
- sampler->min_mip_filter = gl_filter_to_mip_filter(texobj->MinFilter);
- sampler->mag_img_filter = gl_filter_to_img_filter(texobj->MagFilter);
-
- if (texobj->Target != GL_TEXTURE_RECTANGLE_ARB)
- sampler->normalized_coords = 1;
-
- sampler->lod_bias = st->ctx->Texture.Unit[su].LodBias;
-
- sampler->min_lod = texobj->BaseLevel + texobj->MinLod;
- if (sampler->min_lod < texobj->BaseLevel)
- sampler->min_lod = texobj->BaseLevel;
-
- sampler->max_lod = MIN2((GLfloat) texobj->MaxLevel,
- (texobj->MaxLod + texobj->BaseLevel));
- if (sampler->max_lod < sampler->min_lod) {
- /* The GL spec doesn't seem to specify what to do in this case.
- * Swap the values.
- */
- float tmp = sampler->max_lod;
- sampler->max_lod = sampler->min_lod;
- sampler->min_lod = tmp;
- assert(sampler->min_lod <= sampler->max_lod);
- }
-
- xlate_border_color(texobj->BorderColor.f,
- teximg ? teximg->_BaseFormat : GL_RGBA,
- sampler->border_color);
-
- sampler->max_anisotropy = (texobj->MaxAnisotropy == 1.0 ? 0 : (GLuint)texobj->MaxAnisotropy);
-
- /* only care about ARB_shadow, not SGI shadow */
- if (texobj->CompareMode == GL_COMPARE_R_TO_TEXTURE) {
- sampler->compare_mode = PIPE_TEX_COMPARE_R_TO_TEXTURE;
- sampler->compare_func
- = st_compare_func_to_pipe(texobj->CompareFunc);
- }
-
- st->state.num_samplers = su + 1;
-
- /*printf("%s su=%u non-null\n", __FUNCTION__, su);*/
- cso_single_sampler(st->cso_context, su, sampler);
- if (su < st->ctx->Const.MaxVertexTextureImageUnits) {
- cso_single_vertex_sampler(st->cso_context, su, sampler);
- }
- }
- else {
- /*printf("%s su=%u null\n", __FUNCTION__, su);*/
- cso_single_sampler(st->cso_context, su, NULL);
- if (su < st->ctx->Const.MaxVertexTextureImageUnits) {
- cso_single_vertex_sampler(st->cso_context, su, NULL);
- }
- }
- }
-
- cso_single_sampler_done(st->cso_context);
- if (st->ctx->Const.MaxVertexTextureImageUnits > 0) {
- cso_single_vertex_sampler_done(st->cso_context);
- }
-}
-
-
-const struct st_tracked_state st_update_sampler = {
- "st_update_sampler", /* name */
- { /* dirty */
- _NEW_TEXTURE, /* mesa */
- 0, /* st */
- },
- update_samplers /* update */
-};
+/************************************************************************** + * + * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + + /* + * Authors: + * Keith Whitwell <keith@tungstengraphics.com> + * Brian Paul + */ + + +#include "main/macros.h" + +#include "st_context.h" +#include "st_cb_texture.h" +#include "st_format.h" +#include "st_atom.h" +#include "pipe/p_context.h" +#include "pipe/p_defines.h" + +#include "cso_cache/cso_context.h" + + +/** + * Convert GLenum texcoord wrap tokens to pipe tokens. + */ +static GLuint +gl_wrap_xlate(GLenum wrap) +{ + switch (wrap) { + case GL_REPEAT: + return PIPE_TEX_WRAP_REPEAT; + case GL_CLAMP: + return PIPE_TEX_WRAP_CLAMP; + case GL_CLAMP_TO_EDGE: + return PIPE_TEX_WRAP_CLAMP_TO_EDGE; + case GL_CLAMP_TO_BORDER: + return PIPE_TEX_WRAP_CLAMP_TO_BORDER; + case GL_MIRRORED_REPEAT: + return PIPE_TEX_WRAP_MIRROR_REPEAT; + case GL_MIRROR_CLAMP_EXT: + return PIPE_TEX_WRAP_MIRROR_CLAMP; + case GL_MIRROR_CLAMP_TO_EDGE_EXT: + return PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE; + case GL_MIRROR_CLAMP_TO_BORDER_EXT: + return PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER; + default: + assert(0); + return 0; + } +} + + +static GLuint +gl_filter_to_mip_filter(GLenum filter) +{ + switch (filter) { + case GL_NEAREST: + case GL_LINEAR: + return PIPE_TEX_MIPFILTER_NONE; + + case GL_NEAREST_MIPMAP_NEAREST: + case GL_LINEAR_MIPMAP_NEAREST: + return PIPE_TEX_MIPFILTER_NEAREST; + + case GL_NEAREST_MIPMAP_LINEAR: + case GL_LINEAR_MIPMAP_LINEAR: + return PIPE_TEX_MIPFILTER_LINEAR; + + default: + assert(0); + return PIPE_TEX_MIPFILTER_NONE; + } +} + + +static GLuint +gl_filter_to_img_filter(GLenum filter) +{ + switch (filter) { + case GL_NEAREST: + case GL_NEAREST_MIPMAP_NEAREST: + case GL_NEAREST_MIPMAP_LINEAR: + return PIPE_TEX_FILTER_NEAREST; + + case GL_LINEAR: + case GL_LINEAR_MIPMAP_NEAREST: + case GL_LINEAR_MIPMAP_LINEAR: + return PIPE_TEX_FILTER_LINEAR; + + default: + assert(0); + return PIPE_TEX_FILTER_NEAREST; + } +} + + +static void +update_samplers(struct st_context *st) +{ + struct gl_vertex_program *vprog = st->ctx->VertexProgram._Current; + struct gl_fragment_program *fprog = st->ctx->FragmentProgram._Current; + const GLbitfield samplersUsed = (vprog->Base.SamplersUsed | + fprog->Base.SamplersUsed); + GLuint su; + + st->state.num_samplers = 0; + + /* loop over sampler units (aka tex image units) */ + for (su = 0; su < st->ctx->Const.MaxTextureImageUnits; su++) { + struct pipe_sampler_state *sampler = st->state.samplers + su; + + memset(sampler, 0, sizeof(*sampler)); + + if (samplersUsed & (1 << su)) { + struct gl_texture_object *texobj; + struct gl_texture_image *teximg; + GLuint texUnit; + + if (fprog->Base.SamplersUsed & (1 << su)) + texUnit = fprog->Base.SamplerUnits[su]; + else + texUnit = vprog->Base.SamplerUnits[su]; + + texobj = st->ctx->Texture.Unit[texUnit]._Current; + if (!texobj) { + texobj = st_get_default_texture(st); + } + + teximg = texobj->Image[0][texobj->BaseLevel]; + + sampler->wrap_s = gl_wrap_xlate(texobj->WrapS); + sampler->wrap_t = gl_wrap_xlate(texobj->WrapT); + sampler->wrap_r = gl_wrap_xlate(texobj->WrapR); + + sampler->min_img_filter = gl_filter_to_img_filter(texobj->MinFilter); + sampler->min_mip_filter = gl_filter_to_mip_filter(texobj->MinFilter); + sampler->mag_img_filter = gl_filter_to_img_filter(texobj->MagFilter); + + if (texobj->Target != GL_TEXTURE_RECTANGLE_ARB) + sampler->normalized_coords = 1; + + sampler->lod_bias = st->ctx->Texture.Unit[su].LodBias; + + sampler->min_lod = texobj->BaseLevel + texobj->MinLod; + if (sampler->min_lod < texobj->BaseLevel) + sampler->min_lod = texobj->BaseLevel; + + sampler->max_lod = MIN2((GLfloat) texobj->MaxLevel, + (texobj->MaxLod + texobj->BaseLevel)); + if (sampler->max_lod < sampler->min_lod) { + /* The GL spec doesn't seem to specify what to do in this case. + * Swap the values. + */ + float tmp = sampler->max_lod; + sampler->max_lod = sampler->min_lod; + sampler->min_lod = tmp; + assert(sampler->min_lod <= sampler->max_lod); + } + + st_translate_color(texobj->BorderColor.f, + teximg ? teximg->_BaseFormat : GL_RGBA, + sampler->border_color); + + sampler->max_anisotropy = (texobj->MaxAnisotropy == 1.0 ? 0 : (GLuint)texobj->MaxAnisotropy); + + /* only care about ARB_shadow, not SGI shadow */ + if (texobj->CompareMode == GL_COMPARE_R_TO_TEXTURE) { + sampler->compare_mode = PIPE_TEX_COMPARE_R_TO_TEXTURE; + sampler->compare_func + = st_compare_func_to_pipe(texobj->CompareFunc); + } + + st->state.num_samplers = su + 1; + + /*printf("%s su=%u non-null\n", __FUNCTION__, su);*/ + cso_single_sampler(st->cso_context, su, sampler); + if (su < st->ctx->Const.MaxVertexTextureImageUnits) { + cso_single_vertex_sampler(st->cso_context, su, sampler); + } + } + else { + /*printf("%s su=%u null\n", __FUNCTION__, su);*/ + cso_single_sampler(st->cso_context, su, NULL); + if (su < st->ctx->Const.MaxVertexTextureImageUnits) { + cso_single_vertex_sampler(st->cso_context, su, NULL); + } + } + } + + cso_single_sampler_done(st->cso_context); + if (st->ctx->Const.MaxVertexTextureImageUnits > 0) { + cso_single_vertex_sampler_done(st->cso_context); + } +} + + +const struct st_tracked_state st_update_sampler = { + "st_update_sampler", /* name */ + { /* dirty */ + _NEW_TEXTURE, /* mesa */ + 0, /* st */ + }, + update_samplers /* update */ +}; diff --git a/mesalib/src/mesa/state_tracker/st_cb_clear.c b/mesalib/src/mesa/state_tracker/st_cb_clear.c index d2e0cd73c..0e0c4326e 100644 --- a/mesalib/src/mesa/state_tracker/st_cb_clear.c +++ b/mesalib/src/mesa/state_tracker/st_cb_clear.c @@ -42,6 +42,7 @@ #include "st_cb_accum.h" #include "st_cb_clear.h" #include "st_cb_fbo.h" +#include "st_format.h" #include "st_program.h" #include "pipe/p_context.h" @@ -204,6 +205,7 @@ clear_with_quad(struct gl_context *ctx, const GLfloat x1 = (GLfloat) ctx->DrawBuffer->_Xmax / fb_width * 2.0f - 1.0f; const GLfloat y0 = (GLfloat) ctx->DrawBuffer->_Ymin / fb_height * 2.0f - 1.0f; const GLfloat y1 = (GLfloat) ctx->DrawBuffer->_Ymax / fb_height * 2.0f - 1.0f; + float clearColor[4]; /* printf("%s %s%s%s %f,%f %f,%f\n", __FUNCTION__, @@ -298,9 +300,14 @@ clear_with_quad(struct gl_context *ctx, cso_set_fragment_shader_handle(st->cso_context, st->clear.fs); cso_set_vertex_shader_handle(st->cso_context, st->clear.vs); - /* draw quad matching scissor rect (XXX verify coord round-off) */ - draw_quad(st, x0, y0, x1, y1, - (GLfloat) ctx->Depth.Clear, ctx->Color.ClearColor); + if (ctx->DrawBuffer->_ColorDrawBuffers[0]) { + st_translate_color(ctx->Color.ClearColor, + ctx->DrawBuffer->_ColorDrawBuffers[0]->_BaseFormat, + clearColor); + } + + /* draw quad matching scissor rect */ + draw_quad(st, x0, y0, x1, y1, (GLfloat) ctx->Depth.Clear, clearColor); /* Restore pipe state */ cso_restore_blend(st->cso_context); @@ -541,12 +548,21 @@ st_Clear(struct gl_context *ctx, GLbitfield mask) * required from the visual. Hence fix this up to avoid potential * read-modify-write in the driver. */ + float clearColor[4]; + if ((clear_buffers & PIPE_CLEAR_DEPTHSTENCIL) && ((clear_buffers & PIPE_CLEAR_DEPTHSTENCIL) != PIPE_CLEAR_DEPTHSTENCIL) && (depthRb == stencilRb) && (ctx->DrawBuffer->Visual.depthBits == 0 || ctx->DrawBuffer->Visual.stencilBits == 0)) clear_buffers |= PIPE_CLEAR_DEPTHSTENCIL; + + if (ctx->DrawBuffer->_ColorDrawBuffers[0]) { + st_translate_color(ctx->Color.ClearColor, + ctx->DrawBuffer->_ColorDrawBuffers[0]->_BaseFormat, + clearColor); + } + st->pipe->clear(st->pipe, clear_buffers, ctx->Color.ClearColor, ctx->Depth.Clear, ctx->Stencil.Clear); } diff --git a/mesalib/src/mesa/state_tracker/st_cb_texture.c b/mesalib/src/mesa/state_tracker/st_cb_texture.c index c0d1bd94a..a40a79bb8 100644 --- a/mesalib/src/mesa/state_tracker/st_cb_texture.c +++ b/mesalib/src/mesa/state_tracker/st_cb_texture.c @@ -1,1878 +1,1918 @@ -/**************************************************************************
- *
- * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- **************************************************************************/
-
-#include "main/mfeatures.h"
-#include "main/bufferobj.h"
-#include "main/enums.h"
-#include "main/fbobject.h"
-#include "main/formats.h"
-#include "main/image.h"
-#include "main/imports.h"
-#include "main/macros.h"
-#include "main/mipmap.h"
-#include "main/pack.h"
-#include "main/pixeltransfer.h"
-#include "main/texcompress.h"
-#include "main/texfetch.h"
-#include "main/texgetimage.h"
-#include "main/teximage.h"
-#include "main/texobj.h"
-#include "main/texstore.h"
-
-#include "state_tracker/st_debug.h"
-#include "state_tracker/st_context.h"
-#include "state_tracker/st_cb_fbo.h"
-#include "state_tracker/st_cb_flush.h"
-#include "state_tracker/st_cb_texture.h"
-#include "state_tracker/st_format.h"
-#include "state_tracker/st_texture.h"
-#include "state_tracker/st_gen_mipmap.h"
-#include "state_tracker/st_atom.h"
-
-#include "pipe/p_context.h"
-#include "pipe/p_defines.h"
-#include "util/u_inlines.h"
-#include "pipe/p_shader_tokens.h"
-#include "util/u_tile.h"
-#include "util/u_blit.h"
-#include "util/u_format.h"
-#include "util/u_surface.h"
-#include "util/u_sampler.h"
-#include "util/u_math.h"
-#include "util/u_box.h"
-
-#define DBG if (0) printf
-
-
-static enum pipe_texture_target
-gl_target_to_pipe(GLenum target)
-{
- switch (target) {
- case GL_TEXTURE_1D:
- return PIPE_TEXTURE_1D;
- case GL_TEXTURE_2D:
- return PIPE_TEXTURE_2D;
- case GL_TEXTURE_RECTANGLE_NV:
- return PIPE_TEXTURE_RECT;
- case GL_TEXTURE_3D:
- return PIPE_TEXTURE_3D;
- case GL_TEXTURE_CUBE_MAP_ARB:
- return PIPE_TEXTURE_CUBE;
- case GL_TEXTURE_1D_ARRAY_EXT:
- return PIPE_TEXTURE_1D_ARRAY;
- case GL_TEXTURE_2D_ARRAY_EXT:
- return PIPE_TEXTURE_2D_ARRAY;
- default:
- assert(0);
- return 0;
- }
-}
-
-
-/** called via ctx->Driver.NewTextureImage() */
-static struct gl_texture_image *
-st_NewTextureImage(struct gl_context * ctx)
-{
- DBG("%s\n", __FUNCTION__);
- (void) ctx;
- return (struct gl_texture_image *) ST_CALLOC_STRUCT(st_texture_image);
-}
-
-
-/** called via ctx->Driver.NewTextureObject() */
-static struct gl_texture_object *
-st_NewTextureObject(struct gl_context * ctx, GLuint name, GLenum target)
-{
- struct st_texture_object *obj = ST_CALLOC_STRUCT(st_texture_object);
-
- DBG("%s\n", __FUNCTION__);
- _mesa_initialize_texture_object(&obj->base, name, target);
-
- return &obj->base;
-}
-
-/** called via ctx->Driver.DeleteTextureObject() */
-static void
-st_DeleteTextureObject(struct gl_context *ctx,
- struct gl_texture_object *texObj)
-{
- struct st_context *st = st_context(ctx);
- struct st_texture_object *stObj = st_texture_object(texObj);
- if (stObj->pt)
- pipe_resource_reference(&stObj->pt, NULL);
- if (stObj->sampler_view) {
- if (stObj->sampler_view->context != st->pipe) {
- /* Take "ownership" of this texture sampler view by setting
- * its context pointer to this context. This avoids potential
- * crashes when the texture object is shared among contexts
- * and the original/owner context has already been destroyed.
- */
- stObj->sampler_view->context = st->pipe;
- }
- pipe_sampler_view_reference(&stObj->sampler_view, NULL);
- }
- _mesa_delete_texture_object(ctx, texObj);
-}
-
-
-/** called via ctx->Driver.FreeTexImageData() */
-static void
-st_FreeTextureImageData(struct gl_context * ctx, struct gl_texture_image *texImage)
-{
- struct st_texture_image *stImage = st_texture_image(texImage);
-
- DBG("%s\n", __FUNCTION__);
-
- if (stImage->pt) {
- pipe_resource_reference(&stImage->pt, NULL);
- }
-
- if (texImage->Data) {
- _mesa_align_free(texImage->Data);
- texImage->Data = NULL;
- }
-}
-
-
-/**
- * From linux kernel i386 header files, copes with odd sizes better
- * than COPY_DWORDS would:
- * XXX Put this in src/mesa/main/imports.h ???
- */
-#if defined(PIPE_CC_GCC) && defined(PIPE_ARCH_X86)
-static INLINE void *
-__memcpy(void *to, const void *from, size_t n)
-{
- int d0, d1, d2;
- __asm__ __volatile__("rep ; movsl\n\t"
- "testb $2,%b4\n\t"
- "je 1f\n\t"
- "movsw\n"
- "1:\ttestb $1,%b4\n\t"
- "je 2f\n\t"
- "movsb\n" "2:":"=&c"(d0), "=&D"(d1), "=&S"(d2)
- :"0"(n / 4), "q"(n), "1"((long) to), "2"((long) from)
- :"memory");
- return (to);
-}
-#else
-#define __memcpy(a,b,c) memcpy(a,b,c)
-#endif
-
-
-/**
- * The system memcpy (at least on ubuntu 5.10) has problems copying
- * to agp (writecombined) memory from a source which isn't 64-byte
- * aligned - there is a 4x performance falloff.
- *
- * The x86 __memcpy is immune to this but is slightly slower
- * (10%-ish) than the system memcpy.
- *
- * The sse_memcpy seems to have a slight cliff at 64/32 bytes, but
- * isn't much faster than x86_memcpy for agp copies.
- *
- * TODO: switch dynamically.
- */
-static void *
-do_memcpy(void *dest, const void *src, size_t n)
-{
- if ((((unsigned long) src) & 63) || (((unsigned long) dest) & 63)) {
- return __memcpy(dest, src, n);
- }
- else
- return memcpy(dest, src, n);
-}
-
-
-/**
- * Return default texture resource binding bitmask for the given format.
- */
-static GLuint
-default_bindings(struct st_context *st, enum pipe_format format)
-{
- struct pipe_screen *screen = st->pipe->screen;
- const unsigned target = PIPE_TEXTURE_2D;
- const unsigned geom = 0x0;
- unsigned bindings;
-
- if (util_format_is_depth_or_stencil(format))
- bindings = PIPE_BIND_SAMPLER_VIEW | PIPE_BIND_DEPTH_STENCIL;
- else
- bindings = PIPE_BIND_SAMPLER_VIEW | PIPE_BIND_RENDER_TARGET;
-
- if (screen->is_format_supported(screen, format, target, 0, bindings, geom))
- return bindings;
- else
- return PIPE_BIND_SAMPLER_VIEW;
-}
-
-
-/** Return number of image dimensions (1, 2 or 3) for a texture target. */
-static GLuint
-get_texture_dims(GLenum target)
-{
- switch (target) {
- case GL_TEXTURE_1D:
- case GL_TEXTURE_1D_ARRAY_EXT:
- return 1;
- case GL_TEXTURE_2D:
- case GL_TEXTURE_CUBE_MAP_ARB:
- case GL_TEXTURE_CUBE_MAP_POSITIVE_X_ARB:
- case GL_TEXTURE_CUBE_MAP_NEGATIVE_X_ARB:
- case GL_TEXTURE_CUBE_MAP_POSITIVE_Y_ARB:
- case GL_TEXTURE_CUBE_MAP_NEGATIVE_Y_ARB:
- case GL_TEXTURE_CUBE_MAP_POSITIVE_Z_ARB:
- case GL_TEXTURE_CUBE_MAP_NEGATIVE_Z_ARB:
- case GL_TEXTURE_RECTANGLE_NV:
- case GL_TEXTURE_2D_ARRAY_EXT:
- return 2;
- case GL_TEXTURE_3D:
- return 3;
- default:
- assert(0 && "invalid texture target in get_texture_dims()");
- return 1;
- }
-}
-
-
-/**
- * Try to allocate a pipe_resource object for the given st_texture_object.
- *
- * We use the given st_texture_image as a clue to determine the size of the
- * mipmap image at level=0.
- *
- * \return GL_TRUE for success, GL_FALSE if out of memory.
- */
-static GLboolean
-guess_and_alloc_texture(struct st_context *st,
- struct st_texture_object *stObj,
- const struct st_texture_image *stImage)
-{
- const GLuint dims = get_texture_dims(stObj->base.Target);
- GLuint level, lastLevel, width, height, depth;
- GLuint bindings;
- GLuint ptWidth, ptHeight, ptDepth, ptLayers;
- enum pipe_format fmt;
-
- DBG("%s\n", __FUNCTION__);
-
- assert(!stObj->pt);
-
- level = stImage->level;
- width = stImage->base.Width2; /* size w/out border */
- height = stImage->base.Height2;
- depth = stImage->base.Depth2;
-
- assert(width > 0);
- assert(height > 0);
- assert(depth > 0);
-
- /* Depending on the image's size, we can't always make a guess here.
- */
- if (level > 0) {
- if ( (dims >= 1 && width == 1) ||
- (dims >= 2 && height == 1) ||
- (dims >= 3 && depth == 1) ) {
- /* we can't determine the image size at level=0 */
- stObj->width0 = stObj->height0 = stObj->depth0 = 0;
- /* this is not an out of memory error */
- return GL_TRUE;
- }
- }
-
- /* grow the image size until we hit level = 0 */
- while (level > 0) {
- if (width != 1)
- width <<= 1;
- if (height != 1)
- height <<= 1;
- if (depth != 1)
- depth <<= 1;
- level--;
- }
-
- assert(level == 0);
-
- /* At this point, (width x height x depth) is the expected size of
- * the level=0 mipmap image.
- */
-
- /* Guess a reasonable value for lastLevel. With OpenGL we have no
- * idea how many mipmap levels will be in a texture until we start
- * to render with it. Make an educated guess here but be prepared
- * to re-allocating a texture buffer with space for more (or fewer)
- * mipmap levels later.
- */
- if ((stObj->base.MinFilter == GL_NEAREST ||
- stObj->base.MinFilter == GL_LINEAR ||
- stImage->base._BaseFormat == GL_DEPTH_COMPONENT ||
- stImage->base._BaseFormat == GL_DEPTH_STENCIL_EXT) &&
- !stObj->base.GenerateMipmap &&
- stImage->level == 0) {
- /* only alloc space for a single mipmap level */
- lastLevel = 0;
- }
- else {
- /* alloc space for a full mipmap */
- GLuint l2width = util_logbase2(width);
- GLuint l2height = util_logbase2(height);
- GLuint l2depth = util_logbase2(depth);
- lastLevel = MAX2(MAX2(l2width, l2height), l2depth);
- }
-
- /* Save the level=0 dimensions */
- stObj->width0 = width;
- stObj->height0 = height;
- stObj->depth0 = depth;
-
- fmt = st_mesa_format_to_pipe_format(stImage->base.TexFormat);
-
- bindings = default_bindings(st, fmt);
-
- st_gl_texture_dims_to_pipe_dims(stObj->base.Target,
- width, height, depth,
- &ptWidth, &ptHeight, &ptDepth, &ptLayers);
-
- stObj->pt = st_texture_create(st,
- gl_target_to_pipe(stObj->base.Target),
- fmt,
- lastLevel,
- ptWidth,
- ptHeight,
- ptDepth,
- ptLayers,
- bindings);
-
- DBG("%s returning %d\n", __FUNCTION__, (stObj->pt != NULL));
-
- return stObj->pt != NULL;
-}
-
-
-/**
- * Adjust pixel unpack params and image dimensions to strip off the
- * texture border.
- * Gallium doesn't support texture borders. They've seldem been used
- * and seldom been implemented correctly anyway.
- * \param unpackNew returns the new pixel unpack parameters
- */
-static void
-strip_texture_border(GLint border,
- GLint *width, GLint *height, GLint *depth,
- const struct gl_pixelstore_attrib *unpack,
- struct gl_pixelstore_attrib *unpackNew)
-{
- assert(border > 0); /* sanity check */
-
- *unpackNew = *unpack;
-
- if (unpackNew->RowLength == 0)
- unpackNew->RowLength = *width;
-
- if (depth && unpackNew->ImageHeight == 0)
- unpackNew->ImageHeight = *height;
-
- unpackNew->SkipPixels += border;
- if (height)
- unpackNew->SkipRows += border;
- if (depth)
- unpackNew->SkipImages += border;
-
- assert(*width >= 3);
- *width = *width - 2 * border;
- if (height && *height >= 3)
- *height = *height - 2 * border;
- if (depth && *depth >= 3)
- *depth = *depth - 2 * border;
-}
-
-
-/**
- * Do glTexImage1/2/3D().
- */
-static void
-st_TexImage(struct gl_context * ctx,
- GLint dims,
- GLenum target, GLint level,
- GLint internalFormat,
- GLint width, GLint height, GLint depth,
- GLint border,
- GLenum format, GLenum type, const void *pixels,
- const struct gl_pixelstore_attrib *unpack,
- struct gl_texture_object *texObj,
- struct gl_texture_image *texImage,
- GLsizei imageSize, GLboolean compressed_src)
-{
- struct st_context *st = st_context(ctx);
- struct st_texture_object *stObj = st_texture_object(texObj);
- struct st_texture_image *stImage = st_texture_image(texImage);
- GLuint dstRowStride = 0;
- struct gl_pixelstore_attrib unpackNB;
- enum pipe_transfer_usage transfer_usage = 0;
-
- DBG("%s target %s level %d %dx%dx%d border %d\n", __FUNCTION__,
- _mesa_lookup_enum_by_nr(target), level, width, height, depth, border);
-
- /* switch to "normal" */
- if (stObj->surface_based) {
- gl_format texFormat;
-
- _mesa_clear_texture_object(ctx, texObj);
- pipe_resource_reference(&stObj->pt, NULL);
-
- /* oops, need to init this image again */
- texFormat = _mesa_choose_texture_format(ctx, texObj, target, level,
- internalFormat, format, type);
-
- _mesa_init_teximage_fields(ctx, target, texImage,
- width, height, depth, border,
- internalFormat, texFormat);
-
- stObj->surface_based = GL_FALSE;
- }
-
- /* gallium does not support texture borders, strip it off */
- if (border) {
- strip_texture_border(border, &width, &height, &depth, unpack, &unpackNB);
- unpack = &unpackNB;
- texImage->Width = width;
- texImage->Height = height;
- texImage->Depth = depth;
- texImage->Border = 0;
- border = 0;
- }
- else {
- assert(texImage->Width == width);
- assert(texImage->Height == height);
- assert(texImage->Depth == depth);
- }
-
- stImage->face = _mesa_tex_target_to_face(target);
- stImage->level = level;
-
- _mesa_set_fetch_functions(texImage, dims);
-
- /* Release the reference to a potentially orphaned buffer.
- * Release any old malloced memory.
- */
- if (stImage->pt) {
- pipe_resource_reference(&stImage->pt, NULL);
- assert(!texImage->Data);
- }
- else if (texImage->Data) {
- _mesa_align_free(texImage->Data);
- }
-
- /*
- * See if the new image is somehow incompatible with the existing
- * mipmap. If so, free the old mipmap.
- */
- if (stObj->pt) {
- if (level > (GLint) stObj->pt->last_level ||
- !st_texture_match_image(stObj->pt, &stImage->base,
- stImage->face, stImage->level)) {
- DBG("release it\n");
- pipe_resource_reference(&stObj->pt, NULL);
- assert(!stObj->pt);
- pipe_sampler_view_reference(&stObj->sampler_view, NULL);
- }
- }
-
- if (width == 0 || height == 0 || depth == 0) {
- /* stop after freeing old image */
- return;
- }
-
- if (!stObj->pt) {
- if (!guess_and_alloc_texture(st, stObj, stImage)) {
- /* Probably out of memory.
- * Try flushing any pending rendering, then retry.
- */
- st_finish(st);
- if (!guess_and_alloc_texture(st, stObj, stImage)) {
- _mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexImage");
- return;
- }
- }
- }
-
- assert(!stImage->pt);
-
- /* Check if this texture image can live inside the texture object's buffer.
- * If so, store the image there. Otherwise the image will temporarily live
- * in its own buffer.
- */
- if (stObj->pt &&
- st_texture_match_image(stObj->pt, &stImage->base,
- stImage->face, stImage->level)) {
-
- pipe_resource_reference(&stImage->pt, stObj->pt);
- assert(stImage->pt);
- }
-
- if (!stImage->pt)
- DBG("XXX: Image did not fit into texture - storing in local memory!\n");
-
- /* Pixel data may come from regular user memory or a PBO. For the later,
- * do bounds checking and map the PBO to read pixels data from it.
- *
- * XXX we should try to use a GPU-accelerated path to copy the image data
- * from the PBO to the texture.
- */
- if (compressed_src) {
- pixels = _mesa_validate_pbo_compressed_teximage(ctx, imageSize, pixels,
- unpack,
- "glCompressedTexImage");
- }
- else {
- pixels = _mesa_validate_pbo_teximage(ctx, dims, width, height, 1,
- format, type,
- pixels, unpack, "glTexImage");
- }
-
- /*
- * Prepare to store the texture data. Either map the gallium texture buffer
- * memory or malloc space for it.
- */
- if (stImage->pt) {
- /* Store the image in the gallium texture memory buffer */
- if (format == GL_DEPTH_COMPONENT &&
- util_format_is_depth_and_stencil(stImage->pt->format))
- transfer_usage = PIPE_TRANSFER_READ_WRITE;
- else
- transfer_usage = PIPE_TRANSFER_WRITE;
-
- texImage->Data = st_texture_image_map(st, stImage, 0,
- transfer_usage, 0, 0, width, height);
- if(stImage->transfer)
- dstRowStride = stImage->transfer->stride;
- }
- else {
- /* Allocate regular memory and store the image there temporarily. */
- GLuint imageSize = _mesa_format_image_size(texImage->TexFormat,
- width, height, depth);
- dstRowStride = _mesa_format_row_stride(texImage->TexFormat, width);
-
- texImage->Data = _mesa_align_malloc(imageSize, 16);
- }
-
- if (!texImage->Data) {
- _mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexImage");
- return;
- }
-
- if (!pixels) {
- /* We've allocated texture memory, but have no pixel data - all done. */
- goto done;
- }
-
- DBG("Upload image %dx%dx%d row_len %x pitch %x\n",
- width, height, depth, width, dstRowStride);
-
- /* Copy user texture image into the texture buffer.
- */
- if (compressed_src) {
- const GLuint srcRowStride =
- _mesa_format_row_stride(texImage->TexFormat, width);
- if (dstRowStride == srcRowStride) {
- memcpy(texImage->Data, pixels, imageSize);
- }
- else {
- char *dst = texImage->Data;
- const char *src = pixels;
- GLuint i, bw, bh, lines;
- _mesa_get_format_block_size(texImage->TexFormat, &bw, &bh);
- lines = (height + bh - 1) / bh;
-
- for (i = 0; i < lines; ++i) {
- memcpy(dst, src, srcRowStride);
- dst += dstRowStride;
- src += srcRowStride;
- }
- }
- }
- else {
- const GLuint srcImageStride =
- _mesa_image_image_stride(unpack, width, height, format, type);
- GLint i;
- const GLubyte *src = (const GLubyte *) pixels;
-
- for (i = 0; i < depth; i++) {
- if (!_mesa_texstore(ctx, dims,
- texImage->_BaseFormat,
- texImage->TexFormat,
- texImage->Data,
- 0, 0, 0, /* dstX/Y/Zoffset */
- dstRowStride,
- texImage->ImageOffsets,
- width, height, 1,
- format, type, src, unpack)) {
- _mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexImage");
- }
-
- if (stImage->pt && i + 1 < depth) {
- /* unmap this slice */
- st_texture_image_unmap(st, stImage);
- /* map next slice of 3D texture */
- texImage->Data = st_texture_image_map(st, stImage, i + 1,
- transfer_usage, 0, 0,
- width, height);
- src += srcImageStride;
- }
- }
- }
-
-done:
- _mesa_unmap_teximage_pbo(ctx, unpack);
-
- if (stImage->pt && texImage->Data) {
- st_texture_image_unmap(st, stImage);
- texImage->Data = NULL;
- }
-}
-
-
-static void
-st_TexImage3D(struct gl_context * ctx,
- GLenum target, GLint level,
- GLint internalFormat,
- GLint width, GLint height, GLint depth,
- GLint border,
- GLenum format, GLenum type, const void *pixels,
- const struct gl_pixelstore_attrib *unpack,
- struct gl_texture_object *texObj,
- struct gl_texture_image *texImage)
-{
- st_TexImage(ctx, 3, target, level, internalFormat, width, height, depth,
- border, format, type, pixels, unpack, texObj, texImage,
- 0, GL_FALSE);
-}
-
-
-static void
-st_TexImage2D(struct gl_context * ctx,
- GLenum target, GLint level,
- GLint internalFormat,
- GLint width, GLint height, GLint border,
- GLenum format, GLenum type, const void *pixels,
- const struct gl_pixelstore_attrib *unpack,
- struct gl_texture_object *texObj,
- struct gl_texture_image *texImage)
-{
- st_TexImage(ctx, 2, target, level, internalFormat, width, height, 1, border,
- format, type, pixels, unpack, texObj, texImage, 0, GL_FALSE);
-}
-
-
-static void
-st_TexImage1D(struct gl_context * ctx,
- GLenum target, GLint level,
- GLint internalFormat,
- GLint width, GLint border,
- GLenum format, GLenum type, const void *pixels,
- const struct gl_pixelstore_attrib *unpack,
- struct gl_texture_object *texObj,
- struct gl_texture_image *texImage)
-{
- st_TexImage(ctx, 1, target, level, internalFormat, width, 1, 1, border,
- format, type, pixels, unpack, texObj, texImage, 0, GL_FALSE);
-}
-
-
-static void
-st_CompressedTexImage2D(struct gl_context *ctx, GLenum target, GLint level,
- GLint internalFormat,
- GLint width, GLint height, GLint border,
- GLsizei imageSize, const GLvoid *data,
- struct gl_texture_object *texObj,
- struct gl_texture_image *texImage)
-{
- st_TexImage(ctx, 2, target, level, internalFormat, width, height, 1, border,
- 0, 0, data, &ctx->Unpack, texObj, texImage, imageSize, GL_TRUE);
-}
-
-
-
-/**
- * glGetTexImage() helper: decompress a compressed texture by rendering
- * a textured quad. Store the results in the user's buffer.
- */
-static void
-decompress_with_blit(struct gl_context * ctx, GLenum target, GLint level,
- GLenum format, GLenum type, GLvoid *pixels,
- struct gl_texture_object *texObj,
- struct gl_texture_image *texImage)
-{
- struct st_context *st = st_context(ctx);
- struct pipe_context *pipe = st->pipe;
- struct st_texture_image *stImage = st_texture_image(texImage);
- struct st_texture_object *stObj = st_texture_object(texObj);
- struct pipe_sampler_view *src_view =
- st_get_texture_sampler_view(stObj, pipe);
- const GLuint width = texImage->Width;
- const GLuint height = texImage->Height;
- struct pipe_surface *dst_surface;
- struct pipe_resource *dst_texture;
- struct pipe_transfer *tex_xfer;
- unsigned bind = (PIPE_BIND_RENDER_TARGET | /* util_blit may choose to render */
- PIPE_BIND_TRANSFER_READ);
-
- /* create temp / dest surface */
- if (!util_create_rgba_surface(pipe, width, height, bind,
- &dst_texture, &dst_surface)) {
- _mesa_problem(ctx, "util_create_rgba_surface() failed "
- "in decompress_with_blit()");
- return;
- }
-
- /* blit/render/decompress */
- util_blit_pixels_tex(st->blit,
- src_view, /* pipe_resource (src) */
- 0, 0, /* src x0, y0 */
- width, height, /* src x1, y1 */
- dst_surface, /* pipe_surface (dst) */
- 0, 0, /* dst x0, y0 */
- width, height, /* dst x1, y1 */
- 0.0, /* z */
- PIPE_TEX_MIPFILTER_NEAREST);
-
- /* map the dst_surface so we can read from it */
- tex_xfer = pipe_get_transfer(st_context(ctx)->pipe,
- dst_texture, 0, 0,
- PIPE_TRANSFER_READ,
- 0, 0, width, height);
-
- pixels = _mesa_map_pbo_dest(ctx, &ctx->Pack, pixels);
-
- /* copy/pack data into user buffer */
- if (st_equal_formats(stImage->pt->format, format, type)) {
- /* memcpy */
- const uint bytesPerRow = width * util_format_get_blocksize(stImage->pt->format);
- ubyte *map = pipe_transfer_map(pipe, tex_xfer);
- GLuint row;
- for (row = 0; row < height; row++) {
- GLvoid *dest = _mesa_image_address2d(&ctx->Pack, pixels, width,
- height, format, type, row, 0);
- memcpy(dest, map, bytesPerRow);
- map += tex_xfer->stride;
- }
- pipe_transfer_unmap(pipe, tex_xfer);
- }
- else {
- /* format translation via floats */
- GLuint row;
- enum pipe_format format = util_format_linear(dst_texture->format);
- for (row = 0; row < height; row++) {
- const GLbitfield transferOps = 0x0; /* bypassed for glGetTexImage() */
- GLfloat rgba[4 * MAX_WIDTH];
- GLvoid *dest = _mesa_image_address2d(&ctx->Pack, pixels, width,
- height, format, type, row, 0);
-
- if (ST_DEBUG & DEBUG_FALLBACK)
- debug_printf("%s: fallback format translation\n", __FUNCTION__);
-
- /* get float[4] rgba row from surface */
- pipe_get_tile_rgba_format(pipe, tex_xfer, 0, row, width, 1,
- format, rgba);
-
- _mesa_pack_rgba_span_float(ctx, width, (GLfloat (*)[4]) rgba, format,
- type, dest, &ctx->Pack, transferOps);
- }
- }
-
- _mesa_unmap_pbo_dest(ctx, &ctx->Pack);
-
- pipe->transfer_destroy(pipe, tex_xfer);
-
- /* destroy the temp / dest surface */
- util_destroy_rgba_surface(dst_texture, dst_surface);
-}
-
-
-
-/**
- * Need to map texture image into memory before copying image data,
- * then unmap it.
- */
-static void
-st_get_tex_image(struct gl_context * ctx, GLenum target, GLint level,
- GLenum format, GLenum type, GLvoid * pixels,
- struct gl_texture_object *texObj,
- struct gl_texture_image *texImage, GLboolean compressed_dst)
-{
- struct st_context *st = st_context(ctx);
- struct st_texture_image *stImage = st_texture_image(texImage);
- const GLuint dstImageStride =
- _mesa_image_image_stride(&ctx->Pack, texImage->Width, texImage->Height,
- format, type);
- GLuint depth, i;
- GLubyte *dest;
-
- if (stImage->pt &&
- util_format_is_s3tc(stImage->pt->format) &&
- !compressed_dst) {
- /* Need to decompress the texture.
- * We'll do this by rendering a textured quad.
- * Note that we only expect RGBA formats (no Z/depth formats).
- */
- decompress_with_blit(ctx, target, level, format, type, pixels,
- texObj, texImage);
- return;
- }
-
- /* Map */
- if (stImage->pt) {
- /* Image is stored in hardware format in a buffer managed by the
- * kernel. Need to explicitly map and unmap it.
- */
- texImage->Data = st_texture_image_map(st, stImage, 0,
- PIPE_TRANSFER_READ, 0, 0,
- stImage->base.Width,
- stImage->base.Height);
- /* compute stride in texels from stride in bytes */
- texImage->RowStride = stImage->transfer->stride
- * util_format_get_blockwidth(stImage->pt->format)
- / util_format_get_blocksize(stImage->pt->format);
- }
- else {
- /* Otherwise, the image should actually be stored in
- * texImage->Data. This is pretty confusing for
- * everybody, I'd much prefer to separate the two functions of
- * texImage->Data - storage for texture images in main memory
- * and access (ie mappings) of images. In other words, we'd
- * create a new texImage->Map field and leave Data simply for
- * storage.
- */
- assert(texImage->Data);
- }
-
- depth = texImage->Depth;
- texImage->Depth = 1;
-
- dest = (GLubyte *) pixels;
-
- _mesa_set_fetch_functions(texImage, get_texture_dims(target));
-
- for (i = 0; i < depth; i++) {
- if (compressed_dst) {
- _mesa_get_compressed_teximage(ctx, target, level, dest,
- texObj, texImage);
- }
- else {
- _mesa_get_teximage(ctx, target, level, format, type, dest,
- texObj, texImage);
- }
-
- if (stImage->pt && i + 1 < depth) {
- /* unmap this slice */
- st_texture_image_unmap(st, stImage);
- /* map next slice of 3D texture */
- texImage->Data = st_texture_image_map(st, stImage, i + 1,
- PIPE_TRANSFER_READ, 0, 0,
- stImage->base.Width,
- stImage->base.Height);
- dest += dstImageStride;
- }
- }
-
- texImage->Depth = depth;
-
- /* Unmap */
- if (stImage->pt) {
- st_texture_image_unmap(st, stImage);
- texImage->Data = NULL;
- }
-}
-
-
-static void
-st_GetTexImage(struct gl_context * ctx, GLenum target, GLint level,
- GLenum format, GLenum type, GLvoid * pixels,
- struct gl_texture_object *texObj,
- struct gl_texture_image *texImage)
-{
- st_get_tex_image(ctx, target, level, format, type, pixels, texObj, texImage,
- GL_FALSE);
-}
-
-
-static void
-st_GetCompressedTexImage(struct gl_context *ctx, GLenum target, GLint level,
- GLvoid *pixels,
- struct gl_texture_object *texObj,
- struct gl_texture_image *texImage)
-{
- st_get_tex_image(ctx, target, level, 0, 0, pixels, texObj, texImage,
- GL_TRUE);
-}
-
-
-
-static void
-st_TexSubimage(struct gl_context *ctx, GLint dims, GLenum target, GLint level,
- GLint xoffset, GLint yoffset, GLint zoffset,
- GLint width, GLint height, GLint depth,
- GLenum format, GLenum type, const void *pixels,
- const struct gl_pixelstore_attrib *packing,
- struct gl_texture_object *texObj,
- struct gl_texture_image *texImage)
-{
- struct st_context *st = st_context(ctx);
- struct st_texture_image *stImage = st_texture_image(texImage);
- GLuint dstRowStride;
- const GLuint srcImageStride =
- _mesa_image_image_stride(packing, width, height, format, type);
- GLint i;
- const GLubyte *src;
- /* init to silence warning only: */
- enum pipe_transfer_usage transfer_usage = PIPE_TRANSFER_WRITE;
-
- DBG("%s target %s level %d offset %d,%d %dx%d\n", __FUNCTION__,
- _mesa_lookup_enum_by_nr(target),
- level, xoffset, yoffset, width, height);
-
- pixels =
- _mesa_validate_pbo_teximage(ctx, dims, width, height, depth, format,
- type, pixels, packing, "glTexSubImage2D");
- if (!pixels)
- return;
-
- /* Map buffer if necessary. Need to lock to prevent other contexts
- * from uploading the buffer under us.
- */
- if (stImage->pt) {
- if (format == GL_DEPTH_COMPONENT &&
- util_format_is_depth_and_stencil(stImage->pt->format))
- transfer_usage = PIPE_TRANSFER_READ_WRITE;
- else
- transfer_usage = PIPE_TRANSFER_WRITE;
-
- texImage->Data = st_texture_image_map(st, stImage, zoffset,
- transfer_usage,
- xoffset, yoffset,
- width, height);
- }
-
- if (!texImage->Data) {
- _mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexSubImage");
- goto done;
- }
-
- src = (const GLubyte *) pixels;
- dstRowStride = stImage->transfer->stride;
-
- for (i = 0; i < depth; i++) {
- if (!_mesa_texstore(ctx, dims, texImage->_BaseFormat,
- texImage->TexFormat,
- texImage->Data,
- 0, 0, 0,
- dstRowStride,
- texImage->ImageOffsets,
- width, height, 1,
- format, type, src, packing)) {
- _mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexSubImage");
- }
-
- if (stImage->pt && i + 1 < depth) {
- /* unmap this slice */
- st_texture_image_unmap(st, stImage);
- /* map next slice of 3D texture */
- texImage->Data = st_texture_image_map(st, stImage,
- zoffset + i + 1,
- transfer_usage,
- xoffset, yoffset,
- width, height);
- src += srcImageStride;
- }
- }
-
-done:
- _mesa_unmap_teximage_pbo(ctx, packing);
-
- if (stImage->pt && texImage->Data) {
- st_texture_image_unmap(st, stImage);
- texImage->Data = NULL;
- }
-}
-
-
-
-static void
-st_TexSubImage3D(struct gl_context *ctx, GLenum target, GLint level,
- GLint xoffset, GLint yoffset, GLint zoffset,
- GLsizei width, GLsizei height, GLsizei depth,
- GLenum format, GLenum type, const GLvoid *pixels,
- const struct gl_pixelstore_attrib *packing,
- struct gl_texture_object *texObj,
- struct gl_texture_image *texImage)
-{
- st_TexSubimage(ctx, 3, target, level, xoffset, yoffset, zoffset,
- width, height, depth, format, type,
- pixels, packing, texObj, texImage);
-}
-
-
-static void
-st_TexSubImage2D(struct gl_context *ctx, GLenum target, GLint level,
- GLint xoffset, GLint yoffset,
- GLsizei width, GLsizei height,
- GLenum format, GLenum type, const GLvoid * pixels,
- const struct gl_pixelstore_attrib *packing,
- struct gl_texture_object *texObj,
- struct gl_texture_image *texImage)
-{
- st_TexSubimage(ctx, 2, target, level, xoffset, yoffset, 0,
- width, height, 1, format, type,
- pixels, packing, texObj, texImage);
-}
-
-
-static void
-st_TexSubImage1D(struct gl_context *ctx, GLenum target, GLint level,
- GLint xoffset, GLsizei width, GLenum format, GLenum type,
- const GLvoid * pixels,
- const struct gl_pixelstore_attrib *packing,
- struct gl_texture_object *texObj,
- struct gl_texture_image *texImage)
-{
- st_TexSubimage(ctx, 1, target, level, xoffset, 0, 0, width, 1, 1,
- format, type, pixels, packing, texObj, texImage);
-}
-
-
-static void
-st_CompressedTexSubImage1D(struct gl_context *ctx, GLenum target, GLint level,
- GLint xoffset, GLsizei width,
- GLenum format,
- GLsizei imageSize, const GLvoid *data,
- struct gl_texture_object *texObj,
- struct gl_texture_image *texImage)
-{
- assert(0);
-}
-
-
-static void
-st_CompressedTexSubImage2D(struct gl_context *ctx, GLenum target, GLint level,
- GLint xoffset, GLint yoffset,
- GLsizei width, GLint height,
- GLenum format,
- GLsizei imageSize, const GLvoid *data,
- struct gl_texture_object *texObj,
- struct gl_texture_image *texImage)
-{
- struct st_context *st = st_context(ctx);
- struct st_texture_image *stImage = st_texture_image(texImage);
- int srcBlockStride;
- int dstBlockStride;
- int y;
- enum pipe_format pformat;
-
- if (stImage->pt) {
- pformat = stImage->pt->format;
-
- texImage->Data = st_texture_image_map(st, stImage, 0,
- PIPE_TRANSFER_WRITE,
- xoffset, yoffset,
- width, height);
-
- srcBlockStride = util_format_get_stride(pformat, width);
- dstBlockStride = stImage->transfer->stride;
- } else {
- assert(stImage->pt);
- /* TODO find good values for block and strides */
- /* TODO also adjust texImage->data for yoffset/xoffset */
- return;
- }
-
- if (!texImage->Data) {
- _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCompressedTexSubImage");
- return;
- }
-
- assert(xoffset % util_format_get_blockwidth(pformat) == 0);
- assert(yoffset % util_format_get_blockheight(pformat) == 0);
-
- for (y = 0; y < height; y += util_format_get_blockheight(pformat)) {
- /* don't need to adjust for xoffset and yoffset as st_texture_image_map does that */
- const char *src = (const char*)data + srcBlockStride * util_format_get_nblocksy(pformat, y);
- char *dst = (char*)texImage->Data + dstBlockStride * util_format_get_nblocksy(pformat, y);
- memcpy(dst, src, util_format_get_stride(pformat, width));
- }
-
- if (stImage->pt) {
- st_texture_image_unmap(st, stImage);
- texImage->Data = NULL;
- }
-}
-
-
-static void
-st_CompressedTexSubImage3D(struct gl_context *ctx, GLenum target, GLint level,
- GLint xoffset, GLint yoffset, GLint zoffset,
- GLsizei width, GLint height, GLint depth,
- GLenum format,
- GLsizei imageSize, const GLvoid *data,
- struct gl_texture_object *texObj,
- struct gl_texture_image *texImage)
-{
- assert(0);
-}
-
-
-
-/**
- * Do a CopyTexSubImage operation using a read transfer from the source,
- * a write transfer to the destination and get_tile()/put_tile() to access
- * the pixels/texels.
- *
- * Note: srcY=0=TOP of renderbuffer
- */
-static void
-fallback_copy_texsubimage(struct gl_context *ctx, GLenum target, GLint level,
- struct st_renderbuffer *strb,
- struct st_texture_image *stImage,
- GLenum baseFormat,
- GLint destX, GLint destY, GLint destZ,
- GLint srcX, GLint srcY,
- GLsizei width, GLsizei height)
-{
- struct st_context *st = st_context(ctx);
- struct pipe_context *pipe = st->pipe;
- struct pipe_transfer *src_trans;
- GLvoid *texDest;
- enum pipe_transfer_usage transfer_usage;
-
- if (ST_DEBUG & DEBUG_FALLBACK)
- debug_printf("%s: fallback processing\n", __FUNCTION__);
-
- assert(width <= MAX_WIDTH);
-
- if (st_fb_orientation(ctx->ReadBuffer) == Y_0_TOP) {
- srcY = strb->Base.Height - srcY - height;
- }
-
- src_trans = pipe_get_transfer(st_context(ctx)->pipe,
- strb->texture,
- 0, 0,
- PIPE_TRANSFER_READ,
- srcX, srcY,
- width, height);
-
- if ((baseFormat == GL_DEPTH_COMPONENT ||
- baseFormat == GL_DEPTH_STENCIL) &&
- util_format_is_depth_and_stencil(stImage->pt->format))
- transfer_usage = PIPE_TRANSFER_READ_WRITE;
- else
- transfer_usage = PIPE_TRANSFER_WRITE;
-
- /* XXX this used to ignore destZ param */
- texDest = st_texture_image_map(st, stImage, destZ, transfer_usage,
- destX, destY, width, height);
-
- if (baseFormat == GL_DEPTH_COMPONENT ||
- baseFormat == GL_DEPTH_STENCIL) {
- const GLboolean scaleOrBias = (ctx->Pixel.DepthScale != 1.0F ||
- ctx->Pixel.DepthBias != 0.0F);
- GLint row, yStep;
-
- /* determine bottom-to-top vs. top-to-bottom order for src buffer */
- if (st_fb_orientation(ctx->ReadBuffer) == Y_0_TOP) {
- srcY = height - 1;
- yStep = -1;
- }
- else {
- srcY = 0;
- yStep = 1;
- }
-
- /* To avoid a large temp memory allocation, do copy row by row */
- for (row = 0; row < height; row++, srcY += yStep) {
- uint data[MAX_WIDTH];
- pipe_get_tile_z(pipe, src_trans, 0, srcY, width, 1, data);
- if (scaleOrBias) {
- _mesa_scale_and_bias_depth_uint(ctx, width, data);
- }
- pipe_put_tile_z(pipe, stImage->transfer, 0, row, width, 1, data);
- }
- }
- else {
- /* RGBA format */
- GLfloat *tempSrc =
- (GLfloat *) malloc(width * height * 4 * sizeof(GLfloat));
-
- if (tempSrc && texDest) {
- const GLint dims = 2;
- const GLint dstRowStride = stImage->transfer->stride;
- struct gl_texture_image *texImage = &stImage->base;
- struct gl_pixelstore_attrib unpack = ctx->DefaultPacking;
-
- if (st_fb_orientation(ctx->ReadBuffer) == Y_0_TOP) {
- unpack.Invert = GL_TRUE;
- }
-
- /* get float/RGBA image from framebuffer */
- /* XXX this usually involves a lot of int/float conversion.
- * try to avoid that someday.
- */
- pipe_get_tile_rgba_format(pipe, src_trans, 0, 0, width, height,
- util_format_linear(strb->texture->format),
- tempSrc);
-
- /* Store into texture memory.
- * Note that this does some special things such as pixel transfer
- * ops and format conversion. In particular, if the dest tex format
- * is actually RGBA but the user created the texture as GL_RGB we
- * need to fill-in/override the alpha channel with 1.0.
- */
- _mesa_texstore(ctx, dims,
- texImage->_BaseFormat,
- texImage->TexFormat,
- texDest,
- 0, 0, 0,
- dstRowStride,
- texImage->ImageOffsets,
- width, height, 1,
- GL_RGBA, GL_FLOAT, tempSrc, /* src */
- &unpack);
- }
- else {
- _mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexSubImage");
- }
-
- if (tempSrc)
- free(tempSrc);
- }
-
- st_texture_image_unmap(st, stImage);
- pipe->transfer_destroy(pipe, src_trans);
-}
-
-
-
-/**
- * If the format of the src renderbuffer and the format of the dest
- * texture are compatible (in terms of blitting), return a TGSI writemask
- * to be used during the blit.
- * If the src/dest are incompatible, return 0.
- */
-static unsigned
-compatible_src_dst_formats(struct gl_context *ctx,
- const struct gl_renderbuffer *src,
- const struct gl_texture_image *dst)
-{
- /* Get logical base formats for the src and dest.
- * That is, use the user-requested formats and not the actual, device-
- * chosen formats.
- * For example, the user may have requested an A8 texture but the
- * driver may actually be using an RGBA texture format. When we
- * copy/blit to that texture, we only want to copy the Alpha channel
- * and not the RGB channels.
- *
- * Similarly, when the src FBO was created an RGB format may have been
- * requested but the driver actually chose an RGBA format. In that case,
- * we don't want to copy the undefined Alpha channel to the dest texture
- * (it should be 1.0).
- */
- const GLenum srcFormat = _mesa_base_fbo_format(ctx, src->InternalFormat);
- const GLenum dstFormat = _mesa_base_tex_format(ctx, dst->InternalFormat);
-
- /**
- * XXX when we have red-only and red/green renderbuffers we'll need
- * to add more cases here (or implement a general-purpose routine that
- * queries the existance of the R,G,B,A channels in the src and dest).
- */
- if (srcFormat == dstFormat) {
- /* This is the same as matching_base_formats, which should
- * always pass, as it did previously.
- */
- return TGSI_WRITEMASK_XYZW;
- }
- else if (srcFormat == GL_RGB && dstFormat == GL_RGBA) {
- /* Make sure that A in the dest is 1. The actual src format
- * may be RGBA and have undefined A values.
- */
- return TGSI_WRITEMASK_XYZ;
- }
- else if (srcFormat == GL_RGBA && dstFormat == GL_RGB) {
- /* Make sure that A in the dest is 1. The actual dst format
- * may be RGBA and will need A=1 to provide proper alpha values
- * when sampled later.
- */
- return TGSI_WRITEMASK_XYZ;
- }
- else {
- if (ST_DEBUG & DEBUG_FALLBACK)
- debug_printf("%s failed for src %s, dst %s\n",
- __FUNCTION__,
- _mesa_lookup_enum_by_nr(srcFormat),
- _mesa_lookup_enum_by_nr(dstFormat));
-
- /* Otherwise fail.
- */
- return 0;
- }
-}
-
-
-
-/**
- * Do a CopyTex[Sub]Image1/2/3D() using a hardware (blit) path if possible.
- * Note that the region to copy has already been clipped so we know we
- * won't read from outside the source renderbuffer's bounds.
- *
- * Note: srcY=0=Bottom of renderbuffer (GL convention)
- */
-static void
-st_copy_texsubimage(struct gl_context *ctx,
- GLenum target, GLint level,
- GLint destX, GLint destY, GLint destZ,
- GLint srcX, GLint srcY,
- GLsizei width, GLsizei height)
-{
- struct gl_texture_unit *texUnit =
- &ctx->Texture.Unit[ctx->Texture.CurrentUnit];
- struct gl_texture_object *texObj =
- _mesa_select_tex_object(ctx, texUnit, target);
- struct gl_texture_image *texImage =
- _mesa_select_tex_image(ctx, texObj, target, level);
- struct st_texture_image *stImage = st_texture_image(texImage);
- const GLenum texBaseFormat = texImage->_BaseFormat;
- struct gl_framebuffer *fb = ctx->ReadBuffer;
- struct st_renderbuffer *strb;
- struct st_context *st = st_context(ctx);
- struct pipe_context *pipe = st->pipe;
- struct pipe_screen *screen = pipe->screen;
- enum pipe_format dest_format, src_format;
- GLboolean use_fallback = GL_TRUE;
- GLboolean matching_base_formats;
- GLuint format_writemask, sample_count;
- struct pipe_surface *dest_surface = NULL;
- GLboolean do_flip = (st_fb_orientation(ctx->ReadBuffer) == Y_0_TOP);
-
- /* make sure finalize_textures has been called?
- */
- if (0) st_validate_state(st);
-
- /* determine if copying depth or color data */
- if (texBaseFormat == GL_DEPTH_COMPONENT ||
- texBaseFormat == GL_DEPTH_STENCIL) {
- strb = st_renderbuffer(fb->_DepthBuffer);
- if (strb->Base.Wrapped) {
- strb = st_renderbuffer(strb->Base.Wrapped);
- }
- }
- else {
- /* texBaseFormat == GL_RGB, GL_RGBA, GL_ALPHA, etc */
- strb = st_renderbuffer(fb->_ColorReadBuffer);
- }
-
- if (!strb || !strb->surface || !stImage->pt) {
- debug_printf("%s: null strb or stImage\n", __FUNCTION__);
- return;
- }
-
- sample_count = strb->surface->texture->nr_samples;
- /* I believe this would be legal, presumably would need to do a resolve
- for color, and for depth/stencil spec says to just use one of the
- depth/stencil samples per pixel? Need some transfer clarifications. */
- assert(sample_count < 2);
-
- if (srcX < 0) {
- width -= -srcX;
- destX += -srcX;
- srcX = 0;
- }
-
- if (srcY < 0) {
- height -= -srcY;
- destY += -srcY;
- srcY = 0;
- }
-
- if (destX < 0) {
- width -= -destX;
- srcX += -destX;
- destX = 0;
- }
-
- if (destY < 0) {
- height -= -destY;
- srcY += -destY;
- destY = 0;
- }
-
- if (width < 0 || height < 0)
- return;
-
-
- assert(strb);
- assert(strb->surface);
- assert(stImage->pt);
-
- src_format = strb->surface->format;
- dest_format = stImage->pt->format;
-
- /*
- * Determine if the src framebuffer and dest texture have the same
- * base format. We need this to detect a case such as the framebuffer
- * being GL_RGBA but the texture being GL_RGB. If the actual hardware
- * texture format stores RGBA we need to set A=1 (overriding the
- * framebuffer's alpha values). We can't do that with the blit or
- * textured-quad paths.
- */
- matching_base_formats =
- (_mesa_get_format_base_format(strb->Base.Format) ==
- _mesa_get_format_base_format(texImage->TexFormat));
- format_writemask = compatible_src_dst_formats(ctx, &strb->Base, texImage);
-
- if (ctx->_ImageTransferState == 0x0) {
-
- if (matching_base_formats &&
- src_format == dest_format &&
- !do_flip)
- {
- /* use surface_copy() / blit */
- struct pipe_box src_box;
- u_box_2d_zslice(srcX, srcY, strb->surface->u.tex.first_layer,
- width, height, &src_box);
-
- /* for resource_copy_region(), y=0=top, always */
- pipe->resource_copy_region(pipe,
- /* dest */
- stImage->pt,
- stImage->level,
- destX, destY, destZ + stImage->face,
- /* src */
- strb->texture,
- strb->surface->u.tex.level,
- &src_box);
- use_fallback = GL_FALSE;
- }
- else if (format_writemask &&
- texBaseFormat != GL_DEPTH_COMPONENT &&
- texBaseFormat != GL_DEPTH_STENCIL &&
- screen->is_format_supported(screen, src_format,
- PIPE_TEXTURE_2D, sample_count,
- PIPE_BIND_SAMPLER_VIEW,
- 0) &&
- screen->is_format_supported(screen, dest_format,
- PIPE_TEXTURE_2D, 0,
- PIPE_BIND_RENDER_TARGET,
- 0)) {
- /* draw textured quad to do the copy */
- GLint srcY0, srcY1;
- struct pipe_surface surf_tmpl;
- memset(&surf_tmpl, 0, sizeof(surf_tmpl));
- surf_tmpl.format = stImage->pt->format;
- surf_tmpl.usage = PIPE_BIND_RENDER_TARGET;
- surf_tmpl.u.tex.level = stImage->level;
- surf_tmpl.u.tex.first_layer = stImage->face + destZ;
- surf_tmpl.u.tex.last_layer = stImage->face + destZ;
-
- dest_surface = pipe->create_surface(pipe, stImage->pt,
- &surf_tmpl);
-
- if (do_flip) {
- srcY1 = strb->Base.Height - srcY - height;
- srcY0 = srcY1 + height;
- }
- else {
- srcY0 = srcY;
- srcY1 = srcY0 + height;
- }
-
- util_blit_pixels_writemask(st->blit,
- strb->texture,
- strb->surface->u.tex.level,
- srcX, srcY0,
- srcX + width, srcY1,
- strb->surface->u.tex.first_layer,
- dest_surface,
- destX, destY,
- destX + width, destY + height,
- 0.0, PIPE_TEX_MIPFILTER_NEAREST,
- format_writemask);
- use_fallback = GL_FALSE;
- }
-
- if (dest_surface)
- pipe_surface_reference(&dest_surface, NULL);
- }
-
- if (use_fallback) {
- /* software fallback */
- fallback_copy_texsubimage(ctx, target, level,
- strb, stImage, texBaseFormat,
- destX, destY, destZ,
- srcX, srcY, width, height);
- }
-}
-
-
-
-static void
-st_CopyTexImage1D(struct gl_context * ctx, GLenum target, GLint level,
- GLenum internalFormat,
- GLint x, GLint y, GLsizei width, GLint border)
-{
- struct gl_texture_unit *texUnit =
- &ctx->Texture.Unit[ctx->Texture.CurrentUnit];
- struct gl_texture_object *texObj =
- _mesa_select_tex_object(ctx, texUnit, target);
- struct gl_texture_image *texImage =
- _mesa_select_tex_image(ctx, texObj, target, level);
-
- /* Setup or redefine the texture object, texture and texture
- * image. Don't populate yet.
- */
- ctx->Driver.TexImage1D(ctx, target, level, internalFormat,
- width, border,
- GL_RGBA, CHAN_TYPE, NULL,
- &ctx->DefaultPacking, texObj, texImage);
-
- st_copy_texsubimage(ctx, target, level,
- 0, 0, 0, /* destX,Y,Z */
- x, y, width, 1); /* src X, Y, size */
-}
-
-
-static void
-st_CopyTexImage2D(struct gl_context * ctx, GLenum target, GLint level,
- GLenum internalFormat,
- GLint x, GLint y, GLsizei width, GLsizei height,
- GLint border)
-{
- struct gl_texture_unit *texUnit =
- &ctx->Texture.Unit[ctx->Texture.CurrentUnit];
- struct gl_texture_object *texObj =
- _mesa_select_tex_object(ctx, texUnit, target);
- struct gl_texture_image *texImage =
- _mesa_select_tex_image(ctx, texObj, target, level);
-
- /* Setup or redefine the texture object, texture and texture
- * image. Don't populate yet.
- */
- ctx->Driver.TexImage2D(ctx, target, level, internalFormat,
- width, height, border,
- GL_RGBA, CHAN_TYPE, NULL,
- &ctx->DefaultPacking, texObj, texImage);
-
- st_copy_texsubimage(ctx, target, level,
- 0, 0, 0, /* destX,Y,Z */
- x, y, width, height); /* src X, Y, size */
-}
-
-
-static void
-st_CopyTexSubImage1D(struct gl_context * ctx, GLenum target, GLint level,
- GLint xoffset, GLint x, GLint y, GLsizei width)
-{
- const GLint yoffset = 0, zoffset = 0;
- const GLsizei height = 1;
- st_copy_texsubimage(ctx, target, level,
- xoffset, yoffset, zoffset, /* destX,Y,Z */
- x, y, width, height); /* src X, Y, size */
-}
-
-
-static void
-st_CopyTexSubImage2D(struct gl_context * ctx, GLenum target, GLint level,
- GLint xoffset, GLint yoffset,
- GLint x, GLint y, GLsizei width, GLsizei height)
-{
- const GLint zoffset = 0;
- st_copy_texsubimage(ctx, target, level,
- xoffset, yoffset, zoffset, /* destX,Y,Z */
- x, y, width, height); /* src X, Y, size */
-}
-
-
-static void
-st_CopyTexSubImage3D(struct gl_context * ctx, GLenum target, GLint level,
- GLint xoffset, GLint yoffset, GLint zoffset,
- GLint x, GLint y, GLsizei width, GLsizei height)
-{
- st_copy_texsubimage(ctx, target, level,
- xoffset, yoffset, zoffset, /* destX,Y,Z */
- x, y, width, height); /* src X, Y, size */
-}
-
-
-/**
- * Copy image data from stImage into the texture object 'stObj' at level
- * 'dstLevel'.
- */
-static void
-copy_image_data_to_texture(struct st_context *st,
- struct st_texture_object *stObj,
- GLuint dstLevel,
- struct st_texture_image *stImage)
-{
- /* debug checks */
- {
- const struct gl_texture_image *dstImage =
- stObj->base.Image[stImage->face][stImage->level];
- assert(dstImage);
- assert(dstImage->Width == stImage->base.Width);
- assert(dstImage->Height == stImage->base.Height);
- assert(dstImage->Depth == stImage->base.Depth);
- }
-
- if (stImage->pt) {
- /* Copy potentially with the blitter:
- */
- st_texture_image_copy(st->pipe,
- stObj->pt, dstLevel, /* dest texture, level */
- stImage->pt, stImage->level, /* src texture, level */
- stImage->face);
-
- pipe_resource_reference(&stImage->pt, NULL);
- }
- else if (stImage->base.Data) {
- st_texture_image_data(st,
- stObj->pt,
- stImage->face,
- dstLevel,
- stImage->base.Data,
- stImage->base.RowStride *
- util_format_get_blocksize(stObj->pt->format),
- stImage->base.RowStride *
- stImage->base.Height *
- util_format_get_blocksize(stObj->pt->format));
- _mesa_align_free(stImage->base.Data);
- stImage->base.Data = NULL;
- }
-
- pipe_resource_reference(&stImage->pt, stObj->pt);
-}
-
-
-/**
- * Called during state validation. When this function is finished,
- * the texture object should be ready for rendering.
- * \return GL_TRUE for success, GL_FALSE for failure (out of mem)
- */
-GLboolean
-st_finalize_texture(struct gl_context *ctx,
- struct pipe_context *pipe,
- struct gl_texture_object *tObj)
-{
- struct st_context *st = st_context(ctx);
- struct st_texture_object *stObj = st_texture_object(tObj);
- const GLuint nr_faces = (stObj->base.Target == GL_TEXTURE_CUBE_MAP) ? 6 : 1;
- GLuint face;
- struct st_texture_image *firstImage;
- enum pipe_format firstImageFormat;
- GLuint ptWidth, ptHeight, ptDepth, ptLayers;
-
- if (stObj->base._Complete) {
- /* The texture is complete and we know exactly how many mipmap levels
- * are present/needed. This is conditional because we may be called
- * from the st_generate_mipmap() function when the texture object is
- * incomplete. In that case, we'll have set stObj->lastLevel before
- * we get here.
- */
- if (stObj->base.MinFilter == GL_LINEAR ||
- stObj->base.MinFilter == GL_NEAREST)
- stObj->lastLevel = stObj->base.BaseLevel;
- else
- stObj->lastLevel = stObj->base._MaxLevel;
- }
-
- firstImage = st_texture_image(stObj->base.Image[0][stObj->base.BaseLevel]);
- assert(firstImage);
-
- /* If both firstImage and stObj point to a texture which can contain
- * all active images, favour firstImage. Note that because of the
- * completeness requirement, we know that the image dimensions
- * will match.
- */
- if (firstImage->pt &&
- firstImage->pt != stObj->pt &&
- (!stObj->pt || firstImage->pt->last_level >= stObj->pt->last_level)) {
- pipe_resource_reference(&stObj->pt, firstImage->pt);
- pipe_sampler_view_reference(&stObj->sampler_view, NULL);
- }
-
- /* Find gallium format for the Mesa texture */
- firstImageFormat = st_mesa_format_to_pipe_format(firstImage->base.TexFormat);
- st_gl_texture_dims_to_pipe_dims(stObj->base.Target, stObj->width0,
- stObj->height0, stObj->depth0,
- &ptWidth, &ptHeight, &ptDepth, &ptLayers);
-
- /* If we already have a gallium texture, check that it matches the texture
- * object's format, target, size, num_levels, etc.
- */
- if (stObj->pt) {
- if (stObj->pt->target != gl_target_to_pipe(stObj->base.Target) ||
- !st_sampler_compat_formats(stObj->pt->format, firstImageFormat) ||
- stObj->pt->last_level < stObj->lastLevel ||
- stObj->pt->width0 != ptWidth ||
- stObj->pt->height0 != ptHeight ||
- stObj->pt->depth0 != ptDepth ||
- stObj->pt->array_size != ptLayers)
- {
- /* The gallium texture does not match the Mesa texture so delete the
- * gallium texture now. We'll make a new one below.
- */
- pipe_resource_reference(&stObj->pt, NULL);
- pipe_sampler_view_reference(&stObj->sampler_view, NULL);
- st->dirty.st |= ST_NEW_FRAMEBUFFER;
- }
- }
-
- /* May need to create a new gallium texture:
- */
- if (!stObj->pt) {
- GLuint bindings = default_bindings(st, firstImageFormat);
-
- stObj->pt = st_texture_create(st,
- gl_target_to_pipe(stObj->base.Target),
- firstImageFormat,
- stObj->lastLevel,
- ptWidth,
- ptHeight,
- ptDepth,
- ptLayers,
- bindings);
-
- if (!stObj->pt) {
- _mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexImage");
- return GL_FALSE;
- }
- }
-
- /* Pull in any images not in the object's texture:
- */
- for (face = 0; face < nr_faces; face++) {
- GLuint level;
- for (level = stObj->base.BaseLevel; level <= stObj->lastLevel; level++) {
- struct st_texture_image *stImage =
- st_texture_image(stObj->base.Image[face][level]);
-
- /* Need to import images in main memory or held in other textures.
- */
- if (stImage && stObj->pt != stImage->pt) {
- copy_image_data_to_texture(st, stObj, level, stImage);
- }
- }
- }
-
- return GL_TRUE;
-}
-
-
-/**
- * Returns pointer to a default/dummy texture.
- * This is typically used when the current shader has tex/sample instructions
- * but the user has not provided a (any) texture(s).
- */
-struct gl_texture_object *
-st_get_default_texture(struct st_context *st)
-{
- if (!st->default_texture) {
- static const GLenum target = GL_TEXTURE_2D;
- GLubyte pixels[16][16][4];
- struct gl_texture_object *texObj;
- struct gl_texture_image *texImg;
- GLuint i, j;
-
- /* The ARB_fragment_program spec says (0,0,0,1) should be returned
- * when attempting to sample incomplete textures.
- */
- for (i = 0; i < 16; i++) {
- for (j = 0; j < 16; j++) {
- pixels[i][j][0] = 0;
- pixels[i][j][1] = 0;
- pixels[i][j][2] = 0;
- pixels[i][j][3] = 255;
- }
- }
-
- texObj = st->ctx->Driver.NewTextureObject(st->ctx, 0, target);
-
- texImg = _mesa_get_tex_image(st->ctx, texObj, target, 0);
-
- _mesa_init_teximage_fields(st->ctx, target, texImg,
- 16, 16, 1, 0, /* w, h, d, border */
- GL_RGBA, MESA_FORMAT_RGBA8888);
-
- st_TexImage(st->ctx, 2, target,
- 0, GL_RGBA, /* level, intformat */
- 16, 16, 1, 0, /* w, h, d, border */
- GL_RGBA, GL_UNSIGNED_BYTE, pixels,
- &st->ctx->DefaultPacking,
- texObj, texImg,
- 0, 0);
-
- texObj->MinFilter = GL_NEAREST;
- texObj->MagFilter = GL_NEAREST;
- texObj->_Complete = GL_TRUE;
-
- st->default_texture = texObj;
- }
- return st->default_texture;
-}
-
-
-void
-st_init_texture_functions(struct dd_function_table *functions)
-{
- functions->ChooseTextureFormat = st_ChooseTextureFormat;
- functions->TexImage1D = st_TexImage1D;
- functions->TexImage2D = st_TexImage2D;
- functions->TexImage3D = st_TexImage3D;
- functions->TexSubImage1D = st_TexSubImage1D;
- functions->TexSubImage2D = st_TexSubImage2D;
- functions->TexSubImage3D = st_TexSubImage3D;
- functions->CompressedTexSubImage1D = st_CompressedTexSubImage1D;
- functions->CompressedTexSubImage2D = st_CompressedTexSubImage2D;
- functions->CompressedTexSubImage3D = st_CompressedTexSubImage3D;
- functions->CopyTexImage1D = st_CopyTexImage1D;
- functions->CopyTexImage2D = st_CopyTexImage2D;
- functions->CopyTexSubImage1D = st_CopyTexSubImage1D;
- functions->CopyTexSubImage2D = st_CopyTexSubImage2D;
- functions->CopyTexSubImage3D = st_CopyTexSubImage3D;
- functions->GenerateMipmap = st_generate_mipmap;
-
- functions->GetTexImage = st_GetTexImage;
-
- /* compressed texture functions */
- functions->CompressedTexImage2D = st_CompressedTexImage2D;
- functions->GetCompressedTexImage = st_GetCompressedTexImage;
-
- functions->NewTextureObject = st_NewTextureObject;
- functions->NewTextureImage = st_NewTextureImage;
- functions->DeleteTexture = st_DeleteTextureObject;
- functions->FreeTexImageData = st_FreeTextureImageData;
-
- functions->TextureMemCpy = do_memcpy;
-
- /* XXX Temporary until we can query pipe's texture sizes */
- functions->TestProxyTexImage = _mesa_test_proxy_teximage;
-}
+/************************************************************************** + * + * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#include "main/mfeatures.h" +#include "main/bufferobj.h" +#include "main/enums.h" +#include "main/fbobject.h" +#include "main/formats.h" +#include "main/image.h" +#include "main/imports.h" +#include "main/macros.h" +#include "main/mipmap.h" +#include "main/pack.h" +#include "main/pixeltransfer.h" +#include "main/texcompress.h" +#include "main/texfetch.h" +#include "main/texgetimage.h" +#include "main/teximage.h" +#include "main/texobj.h" +#include "main/texstore.h" + +#include "state_tracker/st_debug.h" +#include "state_tracker/st_context.h" +#include "state_tracker/st_cb_fbo.h" +#include "state_tracker/st_cb_flush.h" +#include "state_tracker/st_cb_texture.h" +#include "state_tracker/st_format.h" +#include "state_tracker/st_texture.h" +#include "state_tracker/st_gen_mipmap.h" +#include "state_tracker/st_atom.h" + +#include "pipe/p_context.h" +#include "pipe/p_defines.h" +#include "util/u_inlines.h" +#include "pipe/p_shader_tokens.h" +#include "util/u_tile.h" +#include "util/u_blit.h" +#include "util/u_format.h" +#include "util/u_surface.h" +#include "util/u_sampler.h" +#include "util/u_math.h" +#include "util/u_box.h" + +#define DBG if (0) printf + + +static enum pipe_texture_target +gl_target_to_pipe(GLenum target) +{ + switch (target) { + case GL_TEXTURE_1D: + return PIPE_TEXTURE_1D; + case GL_TEXTURE_2D: + return PIPE_TEXTURE_2D; + case GL_TEXTURE_RECTANGLE_NV: + return PIPE_TEXTURE_RECT; + case GL_TEXTURE_3D: + return PIPE_TEXTURE_3D; + case GL_TEXTURE_CUBE_MAP_ARB: + return PIPE_TEXTURE_CUBE; + case GL_TEXTURE_1D_ARRAY_EXT: + return PIPE_TEXTURE_1D_ARRAY; + case GL_TEXTURE_2D_ARRAY_EXT: + return PIPE_TEXTURE_2D_ARRAY; + default: + assert(0); + return 0; + } +} + + +/** called via ctx->Driver.NewTextureImage() */ +static struct gl_texture_image * +st_NewTextureImage(struct gl_context * ctx) +{ + DBG("%s\n", __FUNCTION__); + (void) ctx; + return (struct gl_texture_image *) ST_CALLOC_STRUCT(st_texture_image); +} + + +/** called via ctx->Driver.NewTextureObject() */ +static struct gl_texture_object * +st_NewTextureObject(struct gl_context * ctx, GLuint name, GLenum target) +{ + struct st_texture_object *obj = ST_CALLOC_STRUCT(st_texture_object); + + DBG("%s\n", __FUNCTION__); + _mesa_initialize_texture_object(&obj->base, name, target); + + return &obj->base; +} + +/** called via ctx->Driver.DeleteTextureObject() */ +static void +st_DeleteTextureObject(struct gl_context *ctx, + struct gl_texture_object *texObj) +{ + struct st_context *st = st_context(ctx); + struct st_texture_object *stObj = st_texture_object(texObj); + if (stObj->pt) + pipe_resource_reference(&stObj->pt, NULL); + if (stObj->sampler_view) { + if (stObj->sampler_view->context != st->pipe) { + /* Take "ownership" of this texture sampler view by setting + * its context pointer to this context. This avoids potential + * crashes when the texture object is shared among contexts + * and the original/owner context has already been destroyed. + */ + stObj->sampler_view->context = st->pipe; + } + pipe_sampler_view_reference(&stObj->sampler_view, NULL); + } + _mesa_delete_texture_object(ctx, texObj); +} + + +/** called via ctx->Driver.FreeTexImageData() */ +static void +st_FreeTextureImageData(struct gl_context * ctx, struct gl_texture_image *texImage) +{ + struct st_texture_image *stImage = st_texture_image(texImage); + + DBG("%s\n", __FUNCTION__); + + if (stImage->pt) { + pipe_resource_reference(&stImage->pt, NULL); + } + + if (texImage->Data) { + _mesa_align_free(texImage->Data); + texImage->Data = NULL; + } +} + + +/** + * From linux kernel i386 header files, copes with odd sizes better + * than COPY_DWORDS would: + * XXX Put this in src/mesa/main/imports.h ??? + */ +#if defined(PIPE_CC_GCC) && defined(PIPE_ARCH_X86) +static INLINE void * +__memcpy(void *to, const void *from, size_t n) +{ + int d0, d1, d2; + __asm__ __volatile__("rep ; movsl\n\t" + "testb $2,%b4\n\t" + "je 1f\n\t" + "movsw\n" + "1:\ttestb $1,%b4\n\t" + "je 2f\n\t" + "movsb\n" "2:":"=&c"(d0), "=&D"(d1), "=&S"(d2) + :"0"(n / 4), "q"(n), "1"((long) to), "2"((long) from) + :"memory"); + return (to); +} +#else +#define __memcpy(a,b,c) memcpy(a,b,c) +#endif + + +/** + * The system memcpy (at least on ubuntu 5.10) has problems copying + * to agp (writecombined) memory from a source which isn't 64-byte + * aligned - there is a 4x performance falloff. + * + * The x86 __memcpy is immune to this but is slightly slower + * (10%-ish) than the system memcpy. + * + * The sse_memcpy seems to have a slight cliff at 64/32 bytes, but + * isn't much faster than x86_memcpy for agp copies. + * + * TODO: switch dynamically. + */ +static void * +do_memcpy(void *dest, const void *src, size_t n) +{ + if ((((unsigned long) src) & 63) || (((unsigned long) dest) & 63)) { + return __memcpy(dest, src, n); + } + else + return memcpy(dest, src, n); +} + + +/** + * Return default texture resource binding bitmask for the given format. + */ +static GLuint +default_bindings(struct st_context *st, enum pipe_format format) +{ + struct pipe_screen *screen = st->pipe->screen; + const unsigned target = PIPE_TEXTURE_2D; + const unsigned geom = 0x0; + unsigned bindings; + + if (util_format_is_depth_or_stencil(format)) + bindings = PIPE_BIND_SAMPLER_VIEW | PIPE_BIND_DEPTH_STENCIL; + else + bindings = PIPE_BIND_SAMPLER_VIEW | PIPE_BIND_RENDER_TARGET; + + if (screen->is_format_supported(screen, format, target, 0, bindings, geom)) + return bindings; + else + return PIPE_BIND_SAMPLER_VIEW; +} + + +/** Return number of image dimensions (1, 2 or 3) for a texture target. */ +static GLuint +get_texture_dims(GLenum target) +{ + switch (target) { + case GL_TEXTURE_1D: + case GL_TEXTURE_1D_ARRAY_EXT: + return 1; + case GL_TEXTURE_2D: + case GL_TEXTURE_CUBE_MAP_ARB: + case GL_TEXTURE_CUBE_MAP_POSITIVE_X_ARB: + case GL_TEXTURE_CUBE_MAP_NEGATIVE_X_ARB: + case GL_TEXTURE_CUBE_MAP_POSITIVE_Y_ARB: + case GL_TEXTURE_CUBE_MAP_NEGATIVE_Y_ARB: + case GL_TEXTURE_CUBE_MAP_POSITIVE_Z_ARB: + case GL_TEXTURE_CUBE_MAP_NEGATIVE_Z_ARB: + case GL_TEXTURE_RECTANGLE_NV: + case GL_TEXTURE_2D_ARRAY_EXT: + return 2; + case GL_TEXTURE_3D: + return 3; + default: + assert(0 && "invalid texture target in get_texture_dims()"); + return 1; + } +} + + +/** + * Given the size of a mipmap image, try to compute the size of the level=0 + * mipmap image. + * + * Note that this isn't always accurate for odd-sized, non-POW textures. + * For example, if level=1 and width=40 then the level=0 width may be 80 or 81. + * + * \return GL_TRUE for success, GL_FALSE for failure + */ +static GLboolean +guess_base_level_size(GLenum target, + GLuint width, GLuint height, GLuint depth, GLuint level, + GLuint *width0, GLuint *height0, GLuint *depth0) +{ + const GLuint dims = get_texture_dims(target); + + assert(width >= 1); + assert(height >= 1); + assert(depth >= 1); + + if (level > 0) { + /* Depending on the image's size, we can't always make a guess here */ + if ((dims >= 1 && width == 1) || + (dims >= 2 && height == 1) || + (dims >= 3 && depth == 1)) { + /* we can't determine the image size at level=0 */ + return GL_FALSE; + } + + /* grow the image size until we hit level = 0 */ + while (level > 0) { + if (width > 1) + width <<= 1; + if (height > 1) + height <<= 1; + if (depth > 1) + depth <<= 1; + level--; + } + } + + *width0 = width; + *height0 = height; + *depth0 = depth; + + return GL_TRUE; +} + + +/** + * Try to allocate a pipe_resource object for the given st_texture_object. + * + * We use the given st_texture_image as a clue to determine the size of the + * mipmap image at level=0. + * + * \return GL_TRUE for success, GL_FALSE if out of memory. + */ +static GLboolean +guess_and_alloc_texture(struct st_context *st, + struct st_texture_object *stObj, + const struct st_texture_image *stImage) +{ + GLuint lastLevel, width, height, depth; + GLuint bindings; + GLuint ptWidth, ptHeight, ptDepth, ptLayers; + enum pipe_format fmt; + + DBG("%s\n", __FUNCTION__); + + assert(!stObj->pt); + + if (!guess_base_level_size(stObj->base.Target, + stImage->base.Width2, + stImage->base.Height2, + stImage->base.Depth2, + stImage->level, + &width, &height, &depth)) { + /* we can't determine the image size at level=0 */ + stObj->width0 = stObj->height0 = stObj->depth0 = 0; + /* this is not an out of memory error */ + return GL_TRUE; + } + + /* At this point, (width x height x depth) is the expected size of + * the level=0 mipmap image. + */ + + /* Guess a reasonable value for lastLevel. With OpenGL we have no + * idea how many mipmap levels will be in a texture until we start + * to render with it. Make an educated guess here but be prepared + * to re-allocating a texture buffer with space for more (or fewer) + * mipmap levels later. + */ + if ((stObj->base.MinFilter == GL_NEAREST || + stObj->base.MinFilter == GL_LINEAR || + stImage->base._BaseFormat == GL_DEPTH_COMPONENT || + stImage->base._BaseFormat == GL_DEPTH_STENCIL_EXT) && + !stObj->base.GenerateMipmap && + stImage->level == 0) { + /* only alloc space for a single mipmap level */ + lastLevel = 0; + } + else { + /* alloc space for a full mipmap */ + GLuint l2width = util_logbase2(width); + GLuint l2height = util_logbase2(height); + GLuint l2depth = util_logbase2(depth); + lastLevel = MAX2(MAX2(l2width, l2height), l2depth); + } + + /* Save the level=0 dimensions */ + stObj->width0 = width; + stObj->height0 = height; + stObj->depth0 = depth; + + fmt = st_mesa_format_to_pipe_format(stImage->base.TexFormat); + + bindings = default_bindings(st, fmt); + + st_gl_texture_dims_to_pipe_dims(stObj->base.Target, + width, height, depth, + &ptWidth, &ptHeight, &ptDepth, &ptLayers); + + stObj->pt = st_texture_create(st, + gl_target_to_pipe(stObj->base.Target), + fmt, + lastLevel, + ptWidth, + ptHeight, + ptDepth, + ptLayers, + bindings); + + DBG("%s returning %d\n", __FUNCTION__, (stObj->pt != NULL)); + + return stObj->pt != NULL; +} + + +/** + * Adjust pixel unpack params and image dimensions to strip off the + * texture border. + * Gallium doesn't support texture borders. They've seldem been used + * and seldom been implemented correctly anyway. + * \param unpackNew returns the new pixel unpack parameters + */ +static void +strip_texture_border(GLint border, + GLint *width, GLint *height, GLint *depth, + const struct gl_pixelstore_attrib *unpack, + struct gl_pixelstore_attrib *unpackNew) +{ + assert(border > 0); /* sanity check */ + + *unpackNew = *unpack; + + if (unpackNew->RowLength == 0) + unpackNew->RowLength = *width; + + if (depth && unpackNew->ImageHeight == 0) + unpackNew->ImageHeight = *height; + + unpackNew->SkipPixels += border; + if (height) + unpackNew->SkipRows += border; + if (depth) + unpackNew->SkipImages += border; + + assert(*width >= 3); + *width = *width - 2 * border; + if (height && *height >= 3) + *height = *height - 2 * border; + if (depth && *depth >= 3) + *depth = *depth - 2 * border; +} + + +/** + * Do glTexImage1/2/3D(). + */ +static void +st_TexImage(struct gl_context * ctx, + GLint dims, + GLenum target, GLint level, + GLint internalFormat, + GLint width, GLint height, GLint depth, + GLint border, + GLenum format, GLenum type, const void *pixels, + const struct gl_pixelstore_attrib *unpack, + struct gl_texture_object *texObj, + struct gl_texture_image *texImage, + GLsizei imageSize, GLboolean compressed_src) +{ + struct st_context *st = st_context(ctx); + struct st_texture_object *stObj = st_texture_object(texObj); + struct st_texture_image *stImage = st_texture_image(texImage); + GLuint dstRowStride = 0; + struct gl_pixelstore_attrib unpackNB; + enum pipe_transfer_usage transfer_usage = 0; + + DBG("%s target %s level %d %dx%dx%d border %d\n", __FUNCTION__, + _mesa_lookup_enum_by_nr(target), level, width, height, depth, border); + + /* switch to "normal" */ + if (stObj->surface_based) { + gl_format texFormat; + + _mesa_clear_texture_object(ctx, texObj); + pipe_resource_reference(&stObj->pt, NULL); + + /* oops, need to init this image again */ + texFormat = _mesa_choose_texture_format(ctx, texObj, target, level, + internalFormat, format, type); + + _mesa_init_teximage_fields(ctx, target, texImage, + width, height, depth, border, + internalFormat, texFormat); + + stObj->surface_based = GL_FALSE; + } + + /* gallium does not support texture borders, strip it off */ + if (border) { + strip_texture_border(border, &width, &height, &depth, unpack, &unpackNB); + unpack = &unpackNB; + texImage->Width = width; + texImage->Height = height; + texImage->Depth = depth; + texImage->Border = 0; + border = 0; + } + else { + assert(texImage->Width == width); + assert(texImage->Height == height); + assert(texImage->Depth == depth); + } + + stImage->face = _mesa_tex_target_to_face(target); + stImage->level = level; + + _mesa_set_fetch_functions(texImage, dims); + + /* Release the reference to a potentially orphaned buffer. + * Release any old malloced memory. + */ + if (stImage->pt) { + pipe_resource_reference(&stImage->pt, NULL); + assert(!texImage->Data); + } + else if (texImage->Data) { + _mesa_align_free(texImage->Data); + } + + /* + * See if the new image is somehow incompatible with the existing + * mipmap. If so, free the old mipmap. + */ + if (stObj->pt) { + if (level > (GLint) stObj->pt->last_level || + !st_texture_match_image(stObj->pt, &stImage->base, + stImage->face, stImage->level)) { + DBG("release it\n"); + pipe_resource_reference(&stObj->pt, NULL); + assert(!stObj->pt); + pipe_sampler_view_reference(&stObj->sampler_view, NULL); + } + } + + if (width == 0 || height == 0 || depth == 0) { + /* stop after freeing old image */ + return; + } + + if (!stObj->pt) { + if (!guess_and_alloc_texture(st, stObj, stImage)) { + /* Probably out of memory. + * Try flushing any pending rendering, then retry. + */ + st_finish(st); + if (!guess_and_alloc_texture(st, stObj, stImage)) { + _mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexImage"); + return; + } + } + } + + assert(!stImage->pt); + + /* Check if this texture image can live inside the texture object's buffer. + * If so, store the image there. Otherwise the image will temporarily live + * in its own buffer. + */ + if (stObj->pt && + st_texture_match_image(stObj->pt, &stImage->base, + stImage->face, stImage->level)) { + + pipe_resource_reference(&stImage->pt, stObj->pt); + assert(stImage->pt); + } + + if (!stImage->pt) + DBG("XXX: Image did not fit into texture - storing in local memory!\n"); + + /* Pixel data may come from regular user memory or a PBO. For the later, + * do bounds checking and map the PBO to read pixels data from it. + * + * XXX we should try to use a GPU-accelerated path to copy the image data + * from the PBO to the texture. + */ + if (compressed_src) { + pixels = _mesa_validate_pbo_compressed_teximage(ctx, imageSize, pixels, + unpack, + "glCompressedTexImage"); + } + else { + pixels = _mesa_validate_pbo_teximage(ctx, dims, width, height, 1, + format, type, + pixels, unpack, "glTexImage"); + } + + /* + * Prepare to store the texture data. Either map the gallium texture buffer + * memory or malloc space for it. + */ + if (stImage->pt) { + /* Store the image in the gallium texture memory buffer */ + if (format == GL_DEPTH_COMPONENT && + util_format_is_depth_and_stencil(stImage->pt->format)) + transfer_usage = PIPE_TRANSFER_READ_WRITE; + else + transfer_usage = PIPE_TRANSFER_WRITE; + + texImage->Data = st_texture_image_map(st, stImage, 0, + transfer_usage, 0, 0, width, height); + if(stImage->transfer) + dstRowStride = stImage->transfer->stride; + } + else { + /* Allocate regular memory and store the image there temporarily. */ + GLuint imageSize = _mesa_format_image_size(texImage->TexFormat, + width, height, depth); + dstRowStride = _mesa_format_row_stride(texImage->TexFormat, width); + + texImage->Data = _mesa_align_malloc(imageSize, 16); + } + + if (!texImage->Data) { + _mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexImage"); + return; + } + + if (!pixels) { + /* We've allocated texture memory, but have no pixel data - all done. */ + goto done; + } + + DBG("Upload image %dx%dx%d row_len %x pitch %x\n", + width, height, depth, width, dstRowStride); + + /* Copy user texture image into the texture buffer. + */ + if (compressed_src) { + const GLuint srcRowStride = + _mesa_format_row_stride(texImage->TexFormat, width); + if (dstRowStride == srcRowStride) { + memcpy(texImage->Data, pixels, imageSize); + } + else { + char *dst = texImage->Data; + const char *src = pixels; + GLuint i, bw, bh, lines; + _mesa_get_format_block_size(texImage->TexFormat, &bw, &bh); + lines = (height + bh - 1) / bh; + + for (i = 0; i < lines; ++i) { + memcpy(dst, src, srcRowStride); + dst += dstRowStride; + src += srcRowStride; + } + } + } + else { + const GLuint srcImageStride = + _mesa_image_image_stride(unpack, width, height, format, type); + GLint i; + const GLubyte *src = (const GLubyte *) pixels; + + for (i = 0; i < depth; i++) { + if (!_mesa_texstore(ctx, dims, + texImage->_BaseFormat, + texImage->TexFormat, + texImage->Data, + 0, 0, 0, /* dstX/Y/Zoffset */ + dstRowStride, + texImage->ImageOffsets, + width, height, 1, + format, type, src, unpack)) { + _mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexImage"); + } + + if (stImage->pt && i + 1 < depth) { + /* unmap this slice */ + st_texture_image_unmap(st, stImage); + /* map next slice of 3D texture */ + texImage->Data = st_texture_image_map(st, stImage, i + 1, + transfer_usage, 0, 0, + width, height); + src += srcImageStride; + } + } + } + +done: + _mesa_unmap_teximage_pbo(ctx, unpack); + + if (stImage->pt && texImage->Data) { + st_texture_image_unmap(st, stImage); + texImage->Data = NULL; + } +} + + +static void +st_TexImage3D(struct gl_context * ctx, + GLenum target, GLint level, + GLint internalFormat, + GLint width, GLint height, GLint depth, + GLint border, + GLenum format, GLenum type, const void *pixels, + const struct gl_pixelstore_attrib *unpack, + struct gl_texture_object *texObj, + struct gl_texture_image *texImage) +{ + st_TexImage(ctx, 3, target, level, internalFormat, width, height, depth, + border, format, type, pixels, unpack, texObj, texImage, + 0, GL_FALSE); +} + + +static void +st_TexImage2D(struct gl_context * ctx, + GLenum target, GLint level, + GLint internalFormat, + GLint width, GLint height, GLint border, + GLenum format, GLenum type, const void *pixels, + const struct gl_pixelstore_attrib *unpack, + struct gl_texture_object *texObj, + struct gl_texture_image *texImage) +{ + st_TexImage(ctx, 2, target, level, internalFormat, width, height, 1, border, + format, type, pixels, unpack, texObj, texImage, 0, GL_FALSE); +} + + +static void +st_TexImage1D(struct gl_context * ctx, + GLenum target, GLint level, + GLint internalFormat, + GLint width, GLint border, + GLenum format, GLenum type, const void *pixels, + const struct gl_pixelstore_attrib *unpack, + struct gl_texture_object *texObj, + struct gl_texture_image *texImage) +{ + st_TexImage(ctx, 1, target, level, internalFormat, width, 1, 1, border, + format, type, pixels, unpack, texObj, texImage, 0, GL_FALSE); +} + + +static void +st_CompressedTexImage2D(struct gl_context *ctx, GLenum target, GLint level, + GLint internalFormat, + GLint width, GLint height, GLint border, + GLsizei imageSize, const GLvoid *data, + struct gl_texture_object *texObj, + struct gl_texture_image *texImage) +{ + st_TexImage(ctx, 2, target, level, internalFormat, width, height, 1, border, + 0, 0, data, &ctx->Unpack, texObj, texImage, imageSize, GL_TRUE); +} + + + +/** + * glGetTexImage() helper: decompress a compressed texture by rendering + * a textured quad. Store the results in the user's buffer. + */ +static void +decompress_with_blit(struct gl_context * ctx, GLenum target, GLint level, + GLenum format, GLenum type, GLvoid *pixels, + struct gl_texture_object *texObj, + struct gl_texture_image *texImage) +{ + struct st_context *st = st_context(ctx); + struct pipe_context *pipe = st->pipe; + struct st_texture_image *stImage = st_texture_image(texImage); + struct st_texture_object *stObj = st_texture_object(texObj); + struct pipe_sampler_view *src_view = + st_get_texture_sampler_view(stObj, pipe); + const GLuint width = texImage->Width; + const GLuint height = texImage->Height; + struct pipe_surface *dst_surface; + struct pipe_resource *dst_texture; + struct pipe_transfer *tex_xfer; + unsigned bind = (PIPE_BIND_RENDER_TARGET | /* util_blit may choose to render */ + PIPE_BIND_TRANSFER_READ); + + /* create temp / dest surface */ + if (!util_create_rgba_surface(pipe, width, height, bind, + &dst_texture, &dst_surface)) { + _mesa_problem(ctx, "util_create_rgba_surface() failed " + "in decompress_with_blit()"); + return; + } + + /* blit/render/decompress */ + util_blit_pixels_tex(st->blit, + src_view, /* pipe_resource (src) */ + 0, 0, /* src x0, y0 */ + width, height, /* src x1, y1 */ + dst_surface, /* pipe_surface (dst) */ + 0, 0, /* dst x0, y0 */ + width, height, /* dst x1, y1 */ + 0.0, /* z */ + PIPE_TEX_MIPFILTER_NEAREST); + + /* map the dst_surface so we can read from it */ + tex_xfer = pipe_get_transfer(st_context(ctx)->pipe, + dst_texture, 0, 0, + PIPE_TRANSFER_READ, + 0, 0, width, height); + + pixels = _mesa_map_pbo_dest(ctx, &ctx->Pack, pixels); + + /* copy/pack data into user buffer */ + if (st_equal_formats(stImage->pt->format, format, type)) { + /* memcpy */ + const uint bytesPerRow = width * util_format_get_blocksize(stImage->pt->format); + ubyte *map = pipe_transfer_map(pipe, tex_xfer); + GLuint row; + for (row = 0; row < height; row++) { + GLvoid *dest = _mesa_image_address2d(&ctx->Pack, pixels, width, + height, format, type, row, 0); + memcpy(dest, map, bytesPerRow); + map += tex_xfer->stride; + } + pipe_transfer_unmap(pipe, tex_xfer); + } + else { + /* format translation via floats */ + GLuint row; + enum pipe_format format = util_format_linear(dst_texture->format); + for (row = 0; row < height; row++) { + const GLbitfield transferOps = 0x0; /* bypassed for glGetTexImage() */ + GLfloat rgba[4 * MAX_WIDTH]; + GLvoid *dest = _mesa_image_address2d(&ctx->Pack, pixels, width, + height, format, type, row, 0); + + if (ST_DEBUG & DEBUG_FALLBACK) + debug_printf("%s: fallback format translation\n", __FUNCTION__); + + /* get float[4] rgba row from surface */ + pipe_get_tile_rgba_format(pipe, tex_xfer, 0, row, width, 1, + format, rgba); + + _mesa_pack_rgba_span_float(ctx, width, (GLfloat (*)[4]) rgba, format, + type, dest, &ctx->Pack, transferOps); + } + } + + _mesa_unmap_pbo_dest(ctx, &ctx->Pack); + + pipe->transfer_destroy(pipe, tex_xfer); + + /* destroy the temp / dest surface */ + util_destroy_rgba_surface(dst_texture, dst_surface); +} + + + +/** + * Need to map texture image into memory before copying image data, + * then unmap it. + */ +static void +st_get_tex_image(struct gl_context * ctx, GLenum target, GLint level, + GLenum format, GLenum type, GLvoid * pixels, + struct gl_texture_object *texObj, + struct gl_texture_image *texImage, GLboolean compressed_dst) +{ + struct st_context *st = st_context(ctx); + struct st_texture_image *stImage = st_texture_image(texImage); + const GLuint dstImageStride = + _mesa_image_image_stride(&ctx->Pack, texImage->Width, texImage->Height, + format, type); + GLuint depth, i; + GLubyte *dest; + + if (stImage->pt && + util_format_is_s3tc(stImage->pt->format) && + !compressed_dst) { + /* Need to decompress the texture. + * We'll do this by rendering a textured quad. + * Note that we only expect RGBA formats (no Z/depth formats). + */ + decompress_with_blit(ctx, target, level, format, type, pixels, + texObj, texImage); + return; + } + + /* Map */ + if (stImage->pt) { + /* Image is stored in hardware format in a buffer managed by the + * kernel. Need to explicitly map and unmap it. + */ + texImage->Data = st_texture_image_map(st, stImage, 0, + PIPE_TRANSFER_READ, 0, 0, + stImage->base.Width, + stImage->base.Height); + /* compute stride in texels from stride in bytes */ + texImage->RowStride = stImage->transfer->stride + * util_format_get_blockwidth(stImage->pt->format) + / util_format_get_blocksize(stImage->pt->format); + } + else { + /* Otherwise, the image should actually be stored in + * texImage->Data. This is pretty confusing for + * everybody, I'd much prefer to separate the two functions of + * texImage->Data - storage for texture images in main memory + * and access (ie mappings) of images. In other words, we'd + * create a new texImage->Map field and leave Data simply for + * storage. + */ + assert(texImage->Data); + } + + depth = texImage->Depth; + texImage->Depth = 1; + + dest = (GLubyte *) pixels; + + _mesa_set_fetch_functions(texImage, get_texture_dims(target)); + + for (i = 0; i < depth; i++) { + if (compressed_dst) { + _mesa_get_compressed_teximage(ctx, target, level, dest, + texObj, texImage); + } + else { + _mesa_get_teximage(ctx, target, level, format, type, dest, + texObj, texImage); + } + + if (stImage->pt && i + 1 < depth) { + /* unmap this slice */ + st_texture_image_unmap(st, stImage); + /* map next slice of 3D texture */ + texImage->Data = st_texture_image_map(st, stImage, i + 1, + PIPE_TRANSFER_READ, 0, 0, + stImage->base.Width, + stImage->base.Height); + dest += dstImageStride; + } + } + + texImage->Depth = depth; + + /* Unmap */ + if (stImage->pt) { + st_texture_image_unmap(st, stImage); + texImage->Data = NULL; + } +} + + +static void +st_GetTexImage(struct gl_context * ctx, GLenum target, GLint level, + GLenum format, GLenum type, GLvoid * pixels, + struct gl_texture_object *texObj, + struct gl_texture_image *texImage) +{ + st_get_tex_image(ctx, target, level, format, type, pixels, texObj, texImage, + GL_FALSE); +} + + +static void +st_GetCompressedTexImage(struct gl_context *ctx, GLenum target, GLint level, + GLvoid *pixels, + struct gl_texture_object *texObj, + struct gl_texture_image *texImage) +{ + st_get_tex_image(ctx, target, level, 0, 0, pixels, texObj, texImage, + GL_TRUE); +} + + + +static void +st_TexSubimage(struct gl_context *ctx, GLint dims, GLenum target, GLint level, + GLint xoffset, GLint yoffset, GLint zoffset, + GLint width, GLint height, GLint depth, + GLenum format, GLenum type, const void *pixels, + const struct gl_pixelstore_attrib *packing, + struct gl_texture_object *texObj, + struct gl_texture_image *texImage) +{ + struct st_context *st = st_context(ctx); + struct st_texture_image *stImage = st_texture_image(texImage); + GLuint dstRowStride; + const GLuint srcImageStride = + _mesa_image_image_stride(packing, width, height, format, type); + GLint i; + const GLubyte *src; + /* init to silence warning only: */ + enum pipe_transfer_usage transfer_usage = PIPE_TRANSFER_WRITE; + + DBG("%s target %s level %d offset %d,%d %dx%d\n", __FUNCTION__, + _mesa_lookup_enum_by_nr(target), + level, xoffset, yoffset, width, height); + + pixels = + _mesa_validate_pbo_teximage(ctx, dims, width, height, depth, format, + type, pixels, packing, "glTexSubImage2D"); + if (!pixels) + return; + + /* Map buffer if necessary. Need to lock to prevent other contexts + * from uploading the buffer under us. + */ + if (stImage->pt) { + if (format == GL_DEPTH_COMPONENT && + util_format_is_depth_and_stencil(stImage->pt->format)) + transfer_usage = PIPE_TRANSFER_READ_WRITE; + else + transfer_usage = PIPE_TRANSFER_WRITE; + + texImage->Data = st_texture_image_map(st, stImage, zoffset, + transfer_usage, + xoffset, yoffset, + width, height); + } + + if (!texImage->Data) { + _mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexSubImage"); + goto done; + } + + src = (const GLubyte *) pixels; + dstRowStride = stImage->transfer->stride; + + for (i = 0; i < depth; i++) { + if (!_mesa_texstore(ctx, dims, texImage->_BaseFormat, + texImage->TexFormat, + texImage->Data, + 0, 0, 0, + dstRowStride, + texImage->ImageOffsets, + width, height, 1, + format, type, src, packing)) { + _mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexSubImage"); + } + + if (stImage->pt && i + 1 < depth) { + /* unmap this slice */ + st_texture_image_unmap(st, stImage); + /* map next slice of 3D texture */ + texImage->Data = st_texture_image_map(st, stImage, + zoffset + i + 1, + transfer_usage, + xoffset, yoffset, + width, height); + src += srcImageStride; + } + } + +done: + _mesa_unmap_teximage_pbo(ctx, packing); + + if (stImage->pt && texImage->Data) { + st_texture_image_unmap(st, stImage); + texImage->Data = NULL; + } +} + + + +static void +st_TexSubImage3D(struct gl_context *ctx, GLenum target, GLint level, + GLint xoffset, GLint yoffset, GLint zoffset, + GLsizei width, GLsizei height, GLsizei depth, + GLenum format, GLenum type, const GLvoid *pixels, + const struct gl_pixelstore_attrib *packing, + struct gl_texture_object *texObj, + struct gl_texture_image *texImage) +{ + st_TexSubimage(ctx, 3, target, level, xoffset, yoffset, zoffset, + width, height, depth, format, type, + pixels, packing, texObj, texImage); +} + + +static void +st_TexSubImage2D(struct gl_context *ctx, GLenum target, GLint level, + GLint xoffset, GLint yoffset, + GLsizei width, GLsizei height, + GLenum format, GLenum type, const GLvoid * pixels, + const struct gl_pixelstore_attrib *packing, + struct gl_texture_object *texObj, + struct gl_texture_image *texImage) +{ + st_TexSubimage(ctx, 2, target, level, xoffset, yoffset, 0, + width, height, 1, format, type, + pixels, packing, texObj, texImage); +} + + +static void +st_TexSubImage1D(struct gl_context *ctx, GLenum target, GLint level, + GLint xoffset, GLsizei width, GLenum format, GLenum type, + const GLvoid * pixels, + const struct gl_pixelstore_attrib *packing, + struct gl_texture_object *texObj, + struct gl_texture_image *texImage) +{ + st_TexSubimage(ctx, 1, target, level, xoffset, 0, 0, width, 1, 1, + format, type, pixels, packing, texObj, texImage); +} + + +static void +st_CompressedTexSubImage1D(struct gl_context *ctx, GLenum target, GLint level, + GLint xoffset, GLsizei width, + GLenum format, + GLsizei imageSize, const GLvoid *data, + struct gl_texture_object *texObj, + struct gl_texture_image *texImage) +{ + assert(0); +} + + +static void +st_CompressedTexSubImage2D(struct gl_context *ctx, GLenum target, GLint level, + GLint xoffset, GLint yoffset, + GLsizei width, GLint height, + GLenum format, + GLsizei imageSize, const GLvoid *data, + struct gl_texture_object *texObj, + struct gl_texture_image *texImage) +{ + struct st_context *st = st_context(ctx); + struct st_texture_image *stImage = st_texture_image(texImage); + int srcBlockStride; + int dstBlockStride; + int y; + enum pipe_format pformat; + + if (stImage->pt) { + pformat = stImage->pt->format; + + texImage->Data = st_texture_image_map(st, stImage, 0, + PIPE_TRANSFER_WRITE, + xoffset, yoffset, + width, height); + + srcBlockStride = util_format_get_stride(pformat, width); + dstBlockStride = stImage->transfer->stride; + } else { + assert(stImage->pt); + /* TODO find good values for block and strides */ + /* TODO also adjust texImage->data for yoffset/xoffset */ + return; + } + + if (!texImage->Data) { + _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCompressedTexSubImage"); + return; + } + + assert(xoffset % util_format_get_blockwidth(pformat) == 0); + assert(yoffset % util_format_get_blockheight(pformat) == 0); + + for (y = 0; y < height; y += util_format_get_blockheight(pformat)) { + /* don't need to adjust for xoffset and yoffset as st_texture_image_map does that */ + const char *src = (const char*)data + srcBlockStride * util_format_get_nblocksy(pformat, y); + char *dst = (char*)texImage->Data + dstBlockStride * util_format_get_nblocksy(pformat, y); + memcpy(dst, src, util_format_get_stride(pformat, width)); + } + + if (stImage->pt) { + st_texture_image_unmap(st, stImage); + texImage->Data = NULL; + } +} + + +static void +st_CompressedTexSubImage3D(struct gl_context *ctx, GLenum target, GLint level, + GLint xoffset, GLint yoffset, GLint zoffset, + GLsizei width, GLint height, GLint depth, + GLenum format, + GLsizei imageSize, const GLvoid *data, + struct gl_texture_object *texObj, + struct gl_texture_image *texImage) +{ + assert(0); +} + + + +/** + * Do a CopyTexSubImage operation using a read transfer from the source, + * a write transfer to the destination and get_tile()/put_tile() to access + * the pixels/texels. + * + * Note: srcY=0=TOP of renderbuffer + */ +static void +fallback_copy_texsubimage(struct gl_context *ctx, GLenum target, GLint level, + struct st_renderbuffer *strb, + struct st_texture_image *stImage, + GLenum baseFormat, + GLint destX, GLint destY, GLint destZ, + GLint srcX, GLint srcY, + GLsizei width, GLsizei height) +{ + struct st_context *st = st_context(ctx); + struct pipe_context *pipe = st->pipe; + struct pipe_transfer *src_trans; + GLvoid *texDest; + enum pipe_transfer_usage transfer_usage; + + if (ST_DEBUG & DEBUG_FALLBACK) + debug_printf("%s: fallback processing\n", __FUNCTION__); + + assert(width <= MAX_WIDTH); + + if (st_fb_orientation(ctx->ReadBuffer) == Y_0_TOP) { + srcY = strb->Base.Height - srcY - height; + } + + src_trans = pipe_get_transfer(st_context(ctx)->pipe, + strb->texture, + 0, 0, + PIPE_TRANSFER_READ, + srcX, srcY, + width, height); + + if ((baseFormat == GL_DEPTH_COMPONENT || + baseFormat == GL_DEPTH_STENCIL) && + util_format_is_depth_and_stencil(stImage->pt->format)) + transfer_usage = PIPE_TRANSFER_READ_WRITE; + else + transfer_usage = PIPE_TRANSFER_WRITE; + + /* XXX this used to ignore destZ param */ + texDest = st_texture_image_map(st, stImage, destZ, transfer_usage, + destX, destY, width, height); + + if (baseFormat == GL_DEPTH_COMPONENT || + baseFormat == GL_DEPTH_STENCIL) { + const GLboolean scaleOrBias = (ctx->Pixel.DepthScale != 1.0F || + ctx->Pixel.DepthBias != 0.0F); + GLint row, yStep; + + /* determine bottom-to-top vs. top-to-bottom order for src buffer */ + if (st_fb_orientation(ctx->ReadBuffer) == Y_0_TOP) { + srcY = height - 1; + yStep = -1; + } + else { + srcY = 0; + yStep = 1; + } + + /* To avoid a large temp memory allocation, do copy row by row */ + for (row = 0; row < height; row++, srcY += yStep) { + uint data[MAX_WIDTH]; + pipe_get_tile_z(pipe, src_trans, 0, srcY, width, 1, data); + if (scaleOrBias) { + _mesa_scale_and_bias_depth_uint(ctx, width, data); + } + pipe_put_tile_z(pipe, stImage->transfer, 0, row, width, 1, data); + } + } + else { + /* RGBA format */ + GLfloat *tempSrc = + (GLfloat *) malloc(width * height * 4 * sizeof(GLfloat)); + + if (tempSrc && texDest) { + const GLint dims = 2; + const GLint dstRowStride = stImage->transfer->stride; + struct gl_texture_image *texImage = &stImage->base; + struct gl_pixelstore_attrib unpack = ctx->DefaultPacking; + + if (st_fb_orientation(ctx->ReadBuffer) == Y_0_TOP) { + unpack.Invert = GL_TRUE; + } + + /* get float/RGBA image from framebuffer */ + /* XXX this usually involves a lot of int/float conversion. + * try to avoid that someday. + */ + pipe_get_tile_rgba_format(pipe, src_trans, 0, 0, width, height, + util_format_linear(strb->texture->format), + tempSrc); + + /* Store into texture memory. + * Note that this does some special things such as pixel transfer + * ops and format conversion. In particular, if the dest tex format + * is actually RGBA but the user created the texture as GL_RGB we + * need to fill-in/override the alpha channel with 1.0. + */ + _mesa_texstore(ctx, dims, + texImage->_BaseFormat, + texImage->TexFormat, + texDest, + 0, 0, 0, + dstRowStride, + texImage->ImageOffsets, + width, height, 1, + GL_RGBA, GL_FLOAT, tempSrc, /* src */ + &unpack); + } + else { + _mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexSubImage"); + } + + if (tempSrc) + free(tempSrc); + } + + st_texture_image_unmap(st, stImage); + pipe->transfer_destroy(pipe, src_trans); +} + + + +/** + * If the format of the src renderbuffer and the format of the dest + * texture are compatible (in terms of blitting), return a TGSI writemask + * to be used during the blit. + * If the src/dest are incompatible, return 0. + */ +static unsigned +compatible_src_dst_formats(struct gl_context *ctx, + const struct gl_renderbuffer *src, + const struct gl_texture_image *dst) +{ + /* Get logical base formats for the src and dest. + * That is, use the user-requested formats and not the actual, device- + * chosen formats. + * For example, the user may have requested an A8 texture but the + * driver may actually be using an RGBA texture format. When we + * copy/blit to that texture, we only want to copy the Alpha channel + * and not the RGB channels. + * + * Similarly, when the src FBO was created an RGB format may have been + * requested but the driver actually chose an RGBA format. In that case, + * we don't want to copy the undefined Alpha channel to the dest texture + * (it should be 1.0). + */ + const GLenum srcFormat = _mesa_base_fbo_format(ctx, src->InternalFormat); + const GLenum dstFormat = _mesa_base_tex_format(ctx, dst->InternalFormat); + + /** + * XXX when we have red-only and red/green renderbuffers we'll need + * to add more cases here (or implement a general-purpose routine that + * queries the existance of the R,G,B,A channels in the src and dest). + */ + if (srcFormat == dstFormat) { + /* This is the same as matching_base_formats, which should + * always pass, as it did previously. + */ + return TGSI_WRITEMASK_XYZW; + } + else if (srcFormat == GL_RGB && dstFormat == GL_RGBA) { + /* Make sure that A in the dest is 1. The actual src format + * may be RGBA and have undefined A values. + */ + return TGSI_WRITEMASK_XYZ; + } + else if (srcFormat == GL_RGBA && dstFormat == GL_RGB) { + /* Make sure that A in the dest is 1. The actual dst format + * may be RGBA and will need A=1 to provide proper alpha values + * when sampled later. + */ + return TGSI_WRITEMASK_XYZ; + } + else { + if (ST_DEBUG & DEBUG_FALLBACK) + debug_printf("%s failed for src %s, dst %s\n", + __FUNCTION__, + _mesa_lookup_enum_by_nr(srcFormat), + _mesa_lookup_enum_by_nr(dstFormat)); + + /* Otherwise fail. + */ + return 0; + } +} + + + +/** + * Do a CopyTex[Sub]Image1/2/3D() using a hardware (blit) path if possible. + * Note that the region to copy has already been clipped so we know we + * won't read from outside the source renderbuffer's bounds. + * + * Note: srcY=0=Bottom of renderbuffer (GL convention) + */ +static void +st_copy_texsubimage(struct gl_context *ctx, + GLenum target, GLint level, + GLint destX, GLint destY, GLint destZ, + GLint srcX, GLint srcY, + GLsizei width, GLsizei height) +{ + struct gl_texture_unit *texUnit = + &ctx->Texture.Unit[ctx->Texture.CurrentUnit]; + struct gl_texture_object *texObj = + _mesa_select_tex_object(ctx, texUnit, target); + struct gl_texture_image *texImage = + _mesa_select_tex_image(ctx, texObj, target, level); + struct st_texture_image *stImage = st_texture_image(texImage); + const GLenum texBaseFormat = texImage->_BaseFormat; + struct gl_framebuffer *fb = ctx->ReadBuffer; + struct st_renderbuffer *strb; + struct st_context *st = st_context(ctx); + struct pipe_context *pipe = st->pipe; + struct pipe_screen *screen = pipe->screen; + enum pipe_format dest_format, src_format; + GLboolean use_fallback = GL_TRUE; + GLboolean matching_base_formats; + GLuint format_writemask, sample_count; + struct pipe_surface *dest_surface = NULL; + GLboolean do_flip = (st_fb_orientation(ctx->ReadBuffer) == Y_0_TOP); + + /* make sure finalize_textures has been called? + */ + if (0) st_validate_state(st); + + /* determine if copying depth or color data */ + if (texBaseFormat == GL_DEPTH_COMPONENT || + texBaseFormat == GL_DEPTH_STENCIL) { + strb = st_renderbuffer(fb->_DepthBuffer); + if (strb->Base.Wrapped) { + strb = st_renderbuffer(strb->Base.Wrapped); + } + } + else { + /* texBaseFormat == GL_RGB, GL_RGBA, GL_ALPHA, etc */ + strb = st_renderbuffer(fb->_ColorReadBuffer); + } + + if (!strb || !strb->surface || !stImage->pt) { + debug_printf("%s: null strb or stImage\n", __FUNCTION__); + return; + } + + sample_count = strb->surface->texture->nr_samples; + /* I believe this would be legal, presumably would need to do a resolve + for color, and for depth/stencil spec says to just use one of the + depth/stencil samples per pixel? Need some transfer clarifications. */ + assert(sample_count < 2); + + if (srcX < 0) { + width -= -srcX; + destX += -srcX; + srcX = 0; + } + + if (srcY < 0) { + height -= -srcY; + destY += -srcY; + srcY = 0; + } + + if (destX < 0) { + width -= -destX; + srcX += -destX; + destX = 0; + } + + if (destY < 0) { + height -= -destY; + srcY += -destY; + destY = 0; + } + + if (width < 0 || height < 0) + return; + + + assert(strb); + assert(strb->surface); + assert(stImage->pt); + + src_format = strb->surface->format; + dest_format = stImage->pt->format; + + /* + * Determine if the src framebuffer and dest texture have the same + * base format. We need this to detect a case such as the framebuffer + * being GL_RGBA but the texture being GL_RGB. If the actual hardware + * texture format stores RGBA we need to set A=1 (overriding the + * framebuffer's alpha values). We can't do that with the blit or + * textured-quad paths. + */ + matching_base_formats = + (_mesa_get_format_base_format(strb->Base.Format) == + _mesa_get_format_base_format(texImage->TexFormat)); + format_writemask = compatible_src_dst_formats(ctx, &strb->Base, texImage); + + if (ctx->_ImageTransferState == 0x0) { + + if (matching_base_formats && + src_format == dest_format && + !do_flip) + { + /* use surface_copy() / blit */ + struct pipe_box src_box; + u_box_2d_zslice(srcX, srcY, strb->surface->u.tex.first_layer, + width, height, &src_box); + + /* for resource_copy_region(), y=0=top, always */ + pipe->resource_copy_region(pipe, + /* dest */ + stImage->pt, + stImage->level, + destX, destY, destZ + stImage->face, + /* src */ + strb->texture, + strb->surface->u.tex.level, + &src_box); + use_fallback = GL_FALSE; + } + else if (format_writemask && + texBaseFormat != GL_DEPTH_COMPONENT && + texBaseFormat != GL_DEPTH_STENCIL && + screen->is_format_supported(screen, src_format, + PIPE_TEXTURE_2D, sample_count, + PIPE_BIND_SAMPLER_VIEW, + 0) && + screen->is_format_supported(screen, dest_format, + PIPE_TEXTURE_2D, 0, + PIPE_BIND_RENDER_TARGET, + 0)) { + /* draw textured quad to do the copy */ + GLint srcY0, srcY1; + struct pipe_surface surf_tmpl; + memset(&surf_tmpl, 0, sizeof(surf_tmpl)); + surf_tmpl.format = stImage->pt->format; + surf_tmpl.usage = PIPE_BIND_RENDER_TARGET; + surf_tmpl.u.tex.level = stImage->level; + surf_tmpl.u.tex.first_layer = stImage->face + destZ; + surf_tmpl.u.tex.last_layer = stImage->face + destZ; + + dest_surface = pipe->create_surface(pipe, stImage->pt, + &surf_tmpl); + + if (do_flip) { + srcY1 = strb->Base.Height - srcY - height; + srcY0 = srcY1 + height; + } + else { + srcY0 = srcY; + srcY1 = srcY0 + height; + } + + util_blit_pixels_writemask(st->blit, + strb->texture, + strb->surface->u.tex.level, + srcX, srcY0, + srcX + width, srcY1, + strb->surface->u.tex.first_layer, + dest_surface, + destX, destY, + destX + width, destY + height, + 0.0, PIPE_TEX_MIPFILTER_NEAREST, + format_writemask); + use_fallback = GL_FALSE; + } + + if (dest_surface) + pipe_surface_reference(&dest_surface, NULL); + } + + if (use_fallback) { + /* software fallback */ + fallback_copy_texsubimage(ctx, target, level, + strb, stImage, texBaseFormat, + destX, destY, destZ, + srcX, srcY, width, height); + } +} + + + +static void +st_CopyTexImage1D(struct gl_context * ctx, GLenum target, GLint level, + GLenum internalFormat, + GLint x, GLint y, GLsizei width, GLint border) +{ + struct gl_texture_unit *texUnit = + &ctx->Texture.Unit[ctx->Texture.CurrentUnit]; + struct gl_texture_object *texObj = + _mesa_select_tex_object(ctx, texUnit, target); + struct gl_texture_image *texImage = + _mesa_select_tex_image(ctx, texObj, target, level); + + /* Setup or redefine the texture object, texture and texture + * image. Don't populate yet. + */ + ctx->Driver.TexImage1D(ctx, target, level, internalFormat, + width, border, + GL_RGBA, CHAN_TYPE, NULL, + &ctx->DefaultPacking, texObj, texImage); + + st_copy_texsubimage(ctx, target, level, + 0, 0, 0, /* destX,Y,Z */ + x, y, width, 1); /* src X, Y, size */ +} + + +static void +st_CopyTexImage2D(struct gl_context * ctx, GLenum target, GLint level, + GLenum internalFormat, + GLint x, GLint y, GLsizei width, GLsizei height, + GLint border) +{ + struct gl_texture_unit *texUnit = + &ctx->Texture.Unit[ctx->Texture.CurrentUnit]; + struct gl_texture_object *texObj = + _mesa_select_tex_object(ctx, texUnit, target); + struct gl_texture_image *texImage = + _mesa_select_tex_image(ctx, texObj, target, level); + + /* Setup or redefine the texture object, texture and texture + * image. Don't populate yet. + */ + ctx->Driver.TexImage2D(ctx, target, level, internalFormat, + width, height, border, + GL_RGBA, CHAN_TYPE, NULL, + &ctx->DefaultPacking, texObj, texImage); + + st_copy_texsubimage(ctx, target, level, + 0, 0, 0, /* destX,Y,Z */ + x, y, width, height); /* src X, Y, size */ +} + + +static void +st_CopyTexSubImage1D(struct gl_context * ctx, GLenum target, GLint level, + GLint xoffset, GLint x, GLint y, GLsizei width) +{ + const GLint yoffset = 0, zoffset = 0; + const GLsizei height = 1; + st_copy_texsubimage(ctx, target, level, + xoffset, yoffset, zoffset, /* destX,Y,Z */ + x, y, width, height); /* src X, Y, size */ +} + + +static void +st_CopyTexSubImage2D(struct gl_context * ctx, GLenum target, GLint level, + GLint xoffset, GLint yoffset, + GLint x, GLint y, GLsizei width, GLsizei height) +{ + const GLint zoffset = 0; + st_copy_texsubimage(ctx, target, level, + xoffset, yoffset, zoffset, /* destX,Y,Z */ + x, y, width, height); /* src X, Y, size */ +} + + +static void +st_CopyTexSubImage3D(struct gl_context * ctx, GLenum target, GLint level, + GLint xoffset, GLint yoffset, GLint zoffset, + GLint x, GLint y, GLsizei width, GLsizei height) +{ + st_copy_texsubimage(ctx, target, level, + xoffset, yoffset, zoffset, /* destX,Y,Z */ + x, y, width, height); /* src X, Y, size */ +} + + +/** + * Copy image data from stImage into the texture object 'stObj' at level + * 'dstLevel'. + */ +static void +copy_image_data_to_texture(struct st_context *st, + struct st_texture_object *stObj, + GLuint dstLevel, + struct st_texture_image *stImage) +{ + /* debug checks */ + { + const struct gl_texture_image *dstImage = + stObj->base.Image[stImage->face][stImage->level]; + assert(dstImage); + assert(dstImage->Width == stImage->base.Width); + assert(dstImage->Height == stImage->base.Height); + assert(dstImage->Depth == stImage->base.Depth); + } + + if (stImage->pt) { + /* Copy potentially with the blitter: + */ + st_texture_image_copy(st->pipe, + stObj->pt, dstLevel, /* dest texture, level */ + stImage->pt, stImage->level, /* src texture, level */ + stImage->face); + + pipe_resource_reference(&stImage->pt, NULL); + } + else if (stImage->base.Data) { + st_texture_image_data(st, + stObj->pt, + stImage->face, + dstLevel, + stImage->base.Data, + stImage->base.RowStride * + util_format_get_blocksize(stObj->pt->format), + stImage->base.RowStride * + stImage->base.Height * + util_format_get_blocksize(stObj->pt->format)); + _mesa_align_free(stImage->base.Data); + stImage->base.Data = NULL; + } + + pipe_resource_reference(&stImage->pt, stObj->pt); +} + + +/** + * Called during state validation. When this function is finished, + * the texture object should be ready for rendering. + * \return GL_TRUE for success, GL_FALSE for failure (out of mem) + */ +GLboolean +st_finalize_texture(struct gl_context *ctx, + struct pipe_context *pipe, + struct gl_texture_object *tObj) +{ + struct st_context *st = st_context(ctx); + struct st_texture_object *stObj = st_texture_object(tObj); + const GLuint nr_faces = (stObj->base.Target == GL_TEXTURE_CUBE_MAP) ? 6 : 1; + GLuint face; + struct st_texture_image *firstImage; + enum pipe_format firstImageFormat; + GLuint ptWidth, ptHeight, ptDepth, ptLayers; + + if (stObj->base._Complete) { + /* The texture is complete and we know exactly how many mipmap levels + * are present/needed. This is conditional because we may be called + * from the st_generate_mipmap() function when the texture object is + * incomplete. In that case, we'll have set stObj->lastLevel before + * we get here. + */ + if (stObj->base.MinFilter == GL_LINEAR || + stObj->base.MinFilter == GL_NEAREST) + stObj->lastLevel = stObj->base.BaseLevel; + else + stObj->lastLevel = stObj->base._MaxLevel; + } + + firstImage = st_texture_image(stObj->base.Image[0][stObj->base.BaseLevel]); + assert(firstImage); + + /* If both firstImage and stObj point to a texture which can contain + * all active images, favour firstImage. Note that because of the + * completeness requirement, we know that the image dimensions + * will match. + */ + if (firstImage->pt && + firstImage->pt != stObj->pt && + (!stObj->pt || firstImage->pt->last_level >= stObj->pt->last_level)) { + pipe_resource_reference(&stObj->pt, firstImage->pt); + pipe_sampler_view_reference(&stObj->sampler_view, NULL); + } + + /* Find gallium format for the Mesa texture */ + firstImageFormat = st_mesa_format_to_pipe_format(firstImage->base.TexFormat); + + /* Find size of level=0 Gallium mipmap image, plus number of texture layers */ + { + GLuint width, height, depth; + if (!guess_base_level_size(stObj->base.Target, + firstImage->base.Width2, + firstImage->base.Height2, + firstImage->base.Depth2, + stObj->base.BaseLevel, + &width, &height, &depth)) { + width = stObj->width0; + height = stObj->height0; + depth = stObj->depth0; + } + /* convert GL dims to Gallium dims */ + st_gl_texture_dims_to_pipe_dims(stObj->base.Target, width, height, depth, + &ptWidth, &ptHeight, &ptDepth, &ptLayers); + } + + /* If we already have a gallium texture, check that it matches the texture + * object's format, target, size, num_levels, etc. + */ + if (stObj->pt) { + if (stObj->pt->target != gl_target_to_pipe(stObj->base.Target) || + !st_sampler_compat_formats(stObj->pt->format, firstImageFormat) || + stObj->pt->last_level < stObj->lastLevel || + stObj->pt->width0 != ptWidth || + stObj->pt->height0 != ptHeight || + stObj->pt->depth0 != ptDepth || + stObj->pt->array_size != ptLayers) + { + /* The gallium texture does not match the Mesa texture so delete the + * gallium texture now. We'll make a new one below. + */ + pipe_resource_reference(&stObj->pt, NULL); + pipe_sampler_view_reference(&stObj->sampler_view, NULL); + st->dirty.st |= ST_NEW_FRAMEBUFFER; + } + } + + /* May need to create a new gallium texture: + */ + if (!stObj->pt) { + GLuint bindings = default_bindings(st, firstImageFormat); + + stObj->pt = st_texture_create(st, + gl_target_to_pipe(stObj->base.Target), + firstImageFormat, + stObj->lastLevel, + ptWidth, + ptHeight, + ptDepth, + ptLayers, + bindings); + + if (!stObj->pt) { + _mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexImage"); + return GL_FALSE; + } + } + + /* Pull in any images not in the object's texture: + */ + for (face = 0; face < nr_faces; face++) { + GLuint level; + for (level = stObj->base.BaseLevel; level <= stObj->lastLevel; level++) { + struct st_texture_image *stImage = + st_texture_image(stObj->base.Image[face][level]); + + /* Need to import images in main memory or held in other textures. + */ + if (stImage && stObj->pt != stImage->pt) { + copy_image_data_to_texture(st, stObj, level, stImage); + } + } + } + + return GL_TRUE; +} + + +/** + * Returns pointer to a default/dummy texture. + * This is typically used when the current shader has tex/sample instructions + * but the user has not provided a (any) texture(s). + */ +struct gl_texture_object * +st_get_default_texture(struct st_context *st) +{ + if (!st->default_texture) { + static const GLenum target = GL_TEXTURE_2D; + GLubyte pixels[16][16][4]; + struct gl_texture_object *texObj; + struct gl_texture_image *texImg; + GLuint i, j; + + /* The ARB_fragment_program spec says (0,0,0,1) should be returned + * when attempting to sample incomplete textures. + */ + for (i = 0; i < 16; i++) { + for (j = 0; j < 16; j++) { + pixels[i][j][0] = 0; + pixels[i][j][1] = 0; + pixels[i][j][2] = 0; + pixels[i][j][3] = 255; + } + } + + texObj = st->ctx->Driver.NewTextureObject(st->ctx, 0, target); + + texImg = _mesa_get_tex_image(st->ctx, texObj, target, 0); + + _mesa_init_teximage_fields(st->ctx, target, texImg, + 16, 16, 1, 0, /* w, h, d, border */ + GL_RGBA, MESA_FORMAT_RGBA8888); + + st_TexImage(st->ctx, 2, target, + 0, GL_RGBA, /* level, intformat */ + 16, 16, 1, 0, /* w, h, d, border */ + GL_RGBA, GL_UNSIGNED_BYTE, pixels, + &st->ctx->DefaultPacking, + texObj, texImg, + 0, 0); + + texObj->MinFilter = GL_NEAREST; + texObj->MagFilter = GL_NEAREST; + texObj->_Complete = GL_TRUE; + + st->default_texture = texObj; + } + return st->default_texture; +} + + +void +st_init_texture_functions(struct dd_function_table *functions) +{ + functions->ChooseTextureFormat = st_ChooseTextureFormat; + functions->TexImage1D = st_TexImage1D; + functions->TexImage2D = st_TexImage2D; + functions->TexImage3D = st_TexImage3D; + functions->TexSubImage1D = st_TexSubImage1D; + functions->TexSubImage2D = st_TexSubImage2D; + functions->TexSubImage3D = st_TexSubImage3D; + functions->CompressedTexSubImage1D = st_CompressedTexSubImage1D; + functions->CompressedTexSubImage2D = st_CompressedTexSubImage2D; + functions->CompressedTexSubImage3D = st_CompressedTexSubImage3D; + functions->CopyTexImage1D = st_CopyTexImage1D; + functions->CopyTexImage2D = st_CopyTexImage2D; + functions->CopyTexSubImage1D = st_CopyTexSubImage1D; + functions->CopyTexSubImage2D = st_CopyTexSubImage2D; + functions->CopyTexSubImage3D = st_CopyTexSubImage3D; + functions->GenerateMipmap = st_generate_mipmap; + + functions->GetTexImage = st_GetTexImage; + + /* compressed texture functions */ + functions->CompressedTexImage2D = st_CompressedTexImage2D; + functions->GetCompressedTexImage = st_GetCompressedTexImage; + + functions->NewTextureObject = st_NewTextureObject; + functions->NewTextureImage = st_NewTextureImage; + functions->DeleteTexture = st_DeleteTextureObject; + functions->FreeTexImageData = st_FreeTextureImageData; + + functions->TextureMemCpy = do_memcpy; + + /* XXX Temporary until we can query pipe's texture sizes */ + functions->TestProxyTexImage = _mesa_test_proxy_teximage; +} diff --git a/mesalib/src/mesa/state_tracker/st_draw.c b/mesalib/src/mesa/state_tracker/st_draw.c index 564d63483..11ebd067e 100644 --- a/mesalib/src/mesa/state_tracker/st_draw.c +++ b/mesalib/src/mesa/state_tracker/st_draw.c @@ -1,751 +1,759 @@ -/**************************************************************************
- *
- * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- **************************************************************************/
-
-/*
- * This file implements the st_draw_vbo() function which is called from
- * Mesa's VBO module. All point/line/triangle rendering is done through
- * this function whether the user called glBegin/End, glDrawArrays,
- * glDrawElements, glEvalMesh, or glCalList, etc.
- *
- * We basically convert the VBO's vertex attribute/array information into
- * Gallium vertex state, bind the vertex buffer objects and call
- * pipe->draw_elements(), pipe->draw_range_elements() or pipe->draw_arrays().
- *
- * Authors:
- * Keith Whitwell <keith@tungstengraphics.com>
- */
-
-
-#include "main/imports.h"
-#include "main/image.h"
-#include "main/macros.h"
-#include "main/mfeatures.h"
-#include "program/prog_uniform.h"
-
-#include "vbo/vbo.h"
-
-#include "st_context.h"
-#include "st_atom.h"
-#include "st_cb_bufferobjects.h"
-#include "st_draw.h"
-#include "st_program.h"
-
-#include "pipe/p_context.h"
-#include "pipe/p_defines.h"
-#include "util/u_inlines.h"
-#include "util/u_format.h"
-#include "util/u_prim.h"
-#include "util/u_draw_quad.h"
-#include "draw/draw_context.h"
-#include "cso_cache/cso_context.h"
-
-
-static GLuint double_types[4] = {
- PIPE_FORMAT_R64_FLOAT,
- PIPE_FORMAT_R64G64_FLOAT,
- PIPE_FORMAT_R64G64B64_FLOAT,
- PIPE_FORMAT_R64G64B64A64_FLOAT
-};
-
-static GLuint float_types[4] = {
- PIPE_FORMAT_R32_FLOAT,
- PIPE_FORMAT_R32G32_FLOAT,
- PIPE_FORMAT_R32G32B32_FLOAT,
- PIPE_FORMAT_R32G32B32A32_FLOAT
-};
-
-static GLuint half_float_types[4] = {
- PIPE_FORMAT_R16_FLOAT,
- PIPE_FORMAT_R16G16_FLOAT,
- PIPE_FORMAT_R16G16B16_FLOAT,
- PIPE_FORMAT_R16G16B16A16_FLOAT
-};
-
-static GLuint uint_types_norm[4] = {
- PIPE_FORMAT_R32_UNORM,
- PIPE_FORMAT_R32G32_UNORM,
- PIPE_FORMAT_R32G32B32_UNORM,
- PIPE_FORMAT_R32G32B32A32_UNORM
-};
-
-static GLuint uint_types_scale[4] = {
- PIPE_FORMAT_R32_USCALED,
- PIPE_FORMAT_R32G32_USCALED,
- PIPE_FORMAT_R32G32B32_USCALED,
- PIPE_FORMAT_R32G32B32A32_USCALED
-};
-
-static GLuint int_types_norm[4] = {
- PIPE_FORMAT_R32_SNORM,
- PIPE_FORMAT_R32G32_SNORM,
- PIPE_FORMAT_R32G32B32_SNORM,
- PIPE_FORMAT_R32G32B32A32_SNORM
-};
-
-static GLuint int_types_scale[4] = {
- PIPE_FORMAT_R32_SSCALED,
- PIPE_FORMAT_R32G32_SSCALED,
- PIPE_FORMAT_R32G32B32_SSCALED,
- PIPE_FORMAT_R32G32B32A32_SSCALED
-};
-
-static GLuint ushort_types_norm[4] = {
- PIPE_FORMAT_R16_UNORM,
- PIPE_FORMAT_R16G16_UNORM,
- PIPE_FORMAT_R16G16B16_UNORM,
- PIPE_FORMAT_R16G16B16A16_UNORM
-};
-
-static GLuint ushort_types_scale[4] = {
- PIPE_FORMAT_R16_USCALED,
- PIPE_FORMAT_R16G16_USCALED,
- PIPE_FORMAT_R16G16B16_USCALED,
- PIPE_FORMAT_R16G16B16A16_USCALED
-};
-
-static GLuint short_types_norm[4] = {
- PIPE_FORMAT_R16_SNORM,
- PIPE_FORMAT_R16G16_SNORM,
- PIPE_FORMAT_R16G16B16_SNORM,
- PIPE_FORMAT_R16G16B16A16_SNORM
-};
-
-static GLuint short_types_scale[4] = {
- PIPE_FORMAT_R16_SSCALED,
- PIPE_FORMAT_R16G16_SSCALED,
- PIPE_FORMAT_R16G16B16_SSCALED,
- PIPE_FORMAT_R16G16B16A16_SSCALED
-};
-
-static GLuint ubyte_types_norm[4] = {
- PIPE_FORMAT_R8_UNORM,
- PIPE_FORMAT_R8G8_UNORM,
- PIPE_FORMAT_R8G8B8_UNORM,
- PIPE_FORMAT_R8G8B8A8_UNORM
-};
-
-static GLuint ubyte_types_scale[4] = {
- PIPE_FORMAT_R8_USCALED,
- PIPE_FORMAT_R8G8_USCALED,
- PIPE_FORMAT_R8G8B8_USCALED,
- PIPE_FORMAT_R8G8B8A8_USCALED
-};
-
-static GLuint byte_types_norm[4] = {
- PIPE_FORMAT_R8_SNORM,
- PIPE_FORMAT_R8G8_SNORM,
- PIPE_FORMAT_R8G8B8_SNORM,
- PIPE_FORMAT_R8G8B8A8_SNORM
-};
-
-static GLuint byte_types_scale[4] = {
- PIPE_FORMAT_R8_SSCALED,
- PIPE_FORMAT_R8G8_SSCALED,
- PIPE_FORMAT_R8G8B8_SSCALED,
- PIPE_FORMAT_R8G8B8A8_SSCALED
-};
-
-static GLuint fixed_types[4] = {
- PIPE_FORMAT_R32_FIXED,
- PIPE_FORMAT_R32G32_FIXED,
- PIPE_FORMAT_R32G32B32_FIXED,
- PIPE_FORMAT_R32G32B32A32_FIXED
-};
-
-
-
-/**
- * Return a PIPE_FORMAT_x for the given GL datatype and size.
- */
-GLuint
-st_pipe_vertex_format(GLenum type, GLuint size, GLenum format,
- GLboolean normalized)
-{
- assert((type >= GL_BYTE && type <= GL_DOUBLE) ||
- type == GL_FIXED || type == GL_HALF_FLOAT);
- assert(size >= 1);
- assert(size <= 4);
- assert(format == GL_RGBA || format == GL_BGRA);
-
- if (format == GL_BGRA) {
- /* this is an odd-ball case */
- assert(type == GL_UNSIGNED_BYTE);
- assert(normalized);
- return PIPE_FORMAT_B8G8R8A8_UNORM;
- }
-
- if (normalized) {
- switch (type) {
- case GL_DOUBLE: return double_types[size-1];
- case GL_FLOAT: return float_types[size-1];
- case GL_HALF_FLOAT: return half_float_types[size-1];
- case GL_INT: return int_types_norm[size-1];
- case GL_SHORT: return short_types_norm[size-1];
- case GL_BYTE: return byte_types_norm[size-1];
- case GL_UNSIGNED_INT: return uint_types_norm[size-1];
- case GL_UNSIGNED_SHORT: return ushort_types_norm[size-1];
- case GL_UNSIGNED_BYTE: return ubyte_types_norm[size-1];
- case GL_FIXED: return fixed_types[size-1];
- default: assert(0); return 0;
- }
- }
- else {
- switch (type) {
- case GL_DOUBLE: return double_types[size-1];
- case GL_FLOAT: return float_types[size-1];
- case GL_HALF_FLOAT: return half_float_types[size-1];
- case GL_INT: return int_types_scale[size-1];
- case GL_SHORT: return short_types_scale[size-1];
- case GL_BYTE: return byte_types_scale[size-1];
- case GL_UNSIGNED_INT: return uint_types_scale[size-1];
- case GL_UNSIGNED_SHORT: return ushort_types_scale[size-1];
- case GL_UNSIGNED_BYTE: return ubyte_types_scale[size-1];
- case GL_FIXED: return fixed_types[size-1];
- default: assert(0); return 0;
- }
- }
- return 0; /* silence compiler warning */
-}
-
-
-
-
-
-/**
- * Examine the active arrays to determine if we have interleaved
- * vertex arrays all living in one VBO, or all living in user space.
- * \param userSpace returns whether the arrays are in user space.
- */
-static GLboolean
-is_interleaved_arrays(const struct st_vertex_program *vp,
- const struct st_vp_variant *vpv,
- const struct gl_client_array **arrays)
-{
- GLuint attr;
- const struct gl_buffer_object *firstBufObj = NULL;
- GLint firstStride = -1;
- const GLubyte *client_addr = NULL;
-
- for (attr = 0; attr < vpv->num_inputs; attr++) {
- const GLuint mesaAttr = vp->index_to_input[attr];
- const struct gl_buffer_object *bufObj = arrays[mesaAttr]->BufferObj;
- const GLsizei stride = arrays[mesaAttr]->StrideB; /* in bytes */
-
- if (firstStride < 0) {
- firstStride = stride;
- }
- else if (firstStride != stride) {
- return GL_FALSE;
- }
-
- if (!bufObj || !bufObj->Name) {
- /* Try to detect if the client-space arrays are
- * "close" to each other.
- */
- if (!client_addr) {
- client_addr = arrays[mesaAttr]->Ptr;
- }
- else if (abs(arrays[mesaAttr]->Ptr - client_addr) > firstStride) {
- /* arrays start too far apart */
- return GL_FALSE;
- }
- }
- else if (!firstBufObj) {
- firstBufObj = bufObj;
- }
- else if (bufObj != firstBufObj) {
- return GL_FALSE;
- }
- }
-
- return GL_TRUE;
-}
-
-
-/**
- * Set up for drawing interleaved arrays that all live in one VBO
- * or all live in user space.
- * \param vbuffer returns vertex buffer info
- * \param velements returns vertex element info
- */
-static void
-setup_interleaved_attribs(struct gl_context *ctx,
- const struct st_vertex_program *vp,
- const struct st_vp_variant *vpv,
- const struct gl_client_array **arrays,
- struct pipe_vertex_buffer *vbuffer,
- struct pipe_vertex_element velements[],
- unsigned max_index)
-{
- struct st_context *st = st_context(ctx);
- struct pipe_context *pipe = st->pipe;
- GLuint attr;
- const GLubyte *low_addr = NULL;
-
- /* Find the lowest address. */
- for (attr = 0; attr < vpv->num_inputs; attr++) {
- const GLubyte *start = arrays[vp->index_to_input[attr]]->Ptr;
-
- low_addr = !low_addr ? start : MIN2(low_addr, start);
- }
-
- for (attr = 0; attr < vpv->num_inputs; attr++) {
- const GLuint mesaAttr = vp->index_to_input[attr];
- struct gl_buffer_object *bufobj = arrays[mesaAttr]->BufferObj;
- struct st_buffer_object *stobj = st_buffer_object(bufobj);
- GLsizei stride = arrays[mesaAttr]->StrideB;
-
- if (attr == 0) {
- if (bufobj && bufobj->Name) {
- vbuffer->buffer = NULL;
- pipe_resource_reference(&vbuffer->buffer, stobj->buffer);
- vbuffer->buffer_offset = pointer_to_offset(low_addr);
- } else {
- vbuffer->buffer =
- pipe_user_buffer_create(pipe->screen, (void*)low_addr,
- stride * (max_index + 1),
- PIPE_BIND_VERTEX_BUFFER);
- vbuffer->buffer_offset = 0;
-
- /* Track user vertex buffers. */
- pipe_resource_reference(&st->user_vb[0], vbuffer->buffer);
- st->user_vb_stride[0] = stride;
- st->num_user_vbs = 1;
- }
- vbuffer->stride = stride; /* in bytes */
- }
-
- velements[attr].src_offset =
- (unsigned) (arrays[mesaAttr]->Ptr - low_addr);
- velements[attr].instance_divisor = arrays[mesaAttr]->InstanceDivisor;
- velements[attr].vertex_buffer_index = 0;
- velements[attr].src_format =
- st_pipe_vertex_format(arrays[mesaAttr]->Type,
- arrays[mesaAttr]->Size,
- arrays[mesaAttr]->Format,
- arrays[mesaAttr]->Normalized);
- assert(velements[attr].src_format);
- }
-}
-
-
-/**
- * Set up a separate pipe_vertex_buffer and pipe_vertex_element for each
- * vertex attribute.
- * \param vbuffer returns vertex buffer info
- * \param velements returns vertex element info
- */
-static void
-setup_non_interleaved_attribs(struct gl_context *ctx,
- const struct st_vertex_program *vp,
- const struct st_vp_variant *vpv,
- const struct gl_client_array **arrays,
- struct pipe_vertex_buffer vbuffer[],
- struct pipe_vertex_element velements[],
- unsigned max_index)
-{
- struct st_context *st = st_context(ctx);
- struct pipe_context *pipe = st->pipe;
- GLuint attr;
-
- for (attr = 0; attr < vpv->num_inputs; attr++) {
- const GLuint mesaAttr = vp->index_to_input[attr];
- struct gl_buffer_object *bufobj = arrays[mesaAttr]->BufferObj;
- GLsizei stride = arrays[mesaAttr]->StrideB;
-
- if (bufobj && bufobj->Name) {
- /* Attribute data is in a VBO.
- * Recall that for VBOs, the gl_client_array->Ptr field is
- * really an offset from the start of the VBO, not a pointer.
- */
- struct st_buffer_object *stobj = st_buffer_object(bufobj);
- assert(stobj->buffer);
-
- vbuffer[attr].buffer = NULL;
- pipe_resource_reference(&vbuffer[attr].buffer, stobj->buffer);
- vbuffer[attr].buffer_offset = pointer_to_offset(arrays[mesaAttr]->Ptr);
- }
- else {
- /* wrap user data */
- if (arrays[mesaAttr]->Ptr) {
- vbuffer[attr].buffer =
- pipe_user_buffer_create(pipe->screen,
- (void *) arrays[mesaAttr]->Ptr,
- stride * (max_index + 1),
- PIPE_BIND_VERTEX_BUFFER);
- }
- else {
- /* no array, use ctx->Current.Attrib[] value */
- uint bytes = sizeof(ctx->Current.Attrib[0]);
- vbuffer[attr].buffer =
- pipe_user_buffer_create(pipe->screen,
- (void *) ctx->Current.Attrib[mesaAttr],
- bytes,
- PIPE_BIND_VERTEX_BUFFER);
- stride = 0;
- }
-
- vbuffer[attr].buffer_offset = 0;
-
- /* Track user vertex buffers. */
- pipe_resource_reference(&st->user_vb[attr], vbuffer->buffer);
- st->user_vb_stride[attr] = stride;
- st->num_user_vbs = MAX2(st->num_user_vbs, attr+1);
- }
-
- /* common-case setup */
- vbuffer[attr].stride = stride; /* in bytes */
-
- velements[attr].src_offset = 0;
- velements[attr].instance_divisor = arrays[mesaAttr]->InstanceDivisor;
- velements[attr].vertex_buffer_index = attr;
- velements[attr].src_format
- = st_pipe_vertex_format(arrays[mesaAttr]->Type,
- arrays[mesaAttr]->Size,
- arrays[mesaAttr]->Format,
- arrays[mesaAttr]->Normalized);
- assert(velements[attr].src_format);
- }
-}
-
-
-static void
-setup_index_buffer(struct gl_context *ctx,
- const struct _mesa_index_buffer *ib,
- struct pipe_index_buffer *ibuffer)
-{
- struct st_context *st = st_context(ctx);
- struct pipe_context *pipe = st->pipe;
-
- memset(ibuffer, 0, sizeof(*ibuffer));
- if (ib) {
- struct gl_buffer_object *bufobj = ib->obj;
-
- switch (ib->type) {
- case GL_UNSIGNED_INT:
- ibuffer->index_size = 4;
- break;
- case GL_UNSIGNED_SHORT:
- ibuffer->index_size = 2;
- break;
- case GL_UNSIGNED_BYTE:
- ibuffer->index_size = 1;
- break;
- default:
- assert(0);
- return;
- }
-
- /* get/create the index buffer object */
- if (bufobj && bufobj->Name) {
- /* elements/indexes are in a real VBO */
- struct st_buffer_object *stobj = st_buffer_object(bufobj);
- pipe_resource_reference(&ibuffer->buffer, stobj->buffer);
- ibuffer->offset = pointer_to_offset(ib->ptr);
- }
- else {
- /* element/indicies are in user space memory */
- ibuffer->buffer =
- pipe_user_buffer_create(pipe->screen, (void *) ib->ptr,
- ib->count * ibuffer->index_size,
- PIPE_BIND_INDEX_BUFFER);
- }
- }
-}
-
-/**
- * Prior to drawing, check that any uniforms referenced by the
- * current shader have been set. If a uniform has not been set,
- * issue a warning.
- */
-static void
-check_uniforms(struct gl_context *ctx)
-{
- struct gl_shader_program *shProg[3] = {
- ctx->Shader.CurrentVertexProgram,
- ctx->Shader.CurrentGeometryProgram,
- ctx->Shader.CurrentFragmentProgram,
- };
- unsigned j;
-
- for (j = 0; j < 3; j++) {
- unsigned i;
-
- if (shProg[j] == NULL || !shProg[j]->LinkStatus)
- continue;
-
- for (i = 0; i < shProg[j]->Uniforms->NumUniforms; i++) {
- const struct gl_uniform *u = &shProg[j]->Uniforms->Uniforms[i];
- if (!u->Initialized) {
- _mesa_warning(ctx,
- "Using shader with uninitialized uniform: %s",
- u->Name);
- }
- }
- }
-}
-
-
-/**
- * Translate OpenGL primtive type (GL_POINTS, GL_TRIANGLE_STRIP, etc) to
- * the corresponding Gallium type.
- */
-static unsigned
-translate_prim(const struct gl_context *ctx, unsigned prim)
-{
- /* GL prims should match Gallium prims, spot-check a few */
- assert(GL_POINTS == PIPE_PRIM_POINTS);
- assert(GL_QUADS == PIPE_PRIM_QUADS);
- assert(GL_TRIANGLE_STRIP_ADJACENCY == PIPE_PRIM_TRIANGLE_STRIP_ADJACENCY);
-
- /* Avoid quadstrips if it's easy to do so:
- * Note: it's imporant to do the correct trimming if we change the prim type!
- * We do that wherever this function is called.
- */
- if (prim == GL_QUAD_STRIP &&
- ctx->Light.ShadeModel != GL_FLAT &&
- ctx->Polygon.FrontMode == GL_FILL &&
- ctx->Polygon.BackMode == GL_FILL)
- prim = GL_TRIANGLE_STRIP;
-
- return prim;
-}
-
-
-static void
-st_validate_varrays(struct gl_context *ctx,
- const struct gl_client_array **arrays,
- unsigned max_index)
-{
- struct st_context *st = st_context(ctx);
- const struct st_vertex_program *vp;
- const struct st_vp_variant *vpv;
- struct pipe_vertex_buffer vbuffer[PIPE_MAX_SHADER_INPUTS];
- struct pipe_vertex_element velements[PIPE_MAX_ATTRIBS];
- unsigned num_vbuffers, num_velements;
- GLuint attr;
- unsigned i;
-
- /* must get these after state validation! */
- vp = st->vp;
- vpv = st->vp_variant;
-
- memset(velements, 0, sizeof(struct pipe_vertex_element) * vpv->num_inputs);
-
- /* Unreference any user vertex buffers. */
- for (i = 0; i < st->num_user_vbs; i++) {
- pipe_resource_reference(&st->user_vb[i], NULL);
- }
- st->num_user_vbs = 0;
-
- /*
- * Setup the vbuffer[] and velements[] arrays.
- */
- if (is_interleaved_arrays(vp, vpv, arrays)) {
- setup_interleaved_attribs(ctx, vp, vpv, arrays, vbuffer, velements,
- max_index);
- num_vbuffers = 1;
- num_velements = vpv->num_inputs;
- if (num_velements == 0)
- num_vbuffers = 0;
- }
- else {
- setup_non_interleaved_attribs(ctx, vp, vpv, arrays,
- vbuffer, velements, max_index);
- num_vbuffers = vpv->num_inputs;
- num_velements = vpv->num_inputs;
- }
-
- cso_set_vertex_buffers(st->cso_context, num_vbuffers, vbuffer);
- cso_set_vertex_elements(st->cso_context, num_velements, velements);
-
- /* unreference buffers (frees wrapped user-space buffer objects)
- * This is OK, because the pipe driver should reference buffers by itself
- * in set_vertex_buffers. */
- for (attr = 0; attr < num_vbuffers; attr++) {
- pipe_resource_reference(&vbuffer[attr].buffer, NULL);
- assert(!vbuffer[attr].buffer);
- }
-}
-
-
-/**
- * This function gets plugged into the VBO module and is called when
- * we have something to render.
- * Basically, translate the information into the format expected by gallium.
- */
-void
-st_draw_vbo(struct gl_context *ctx,
- const struct gl_client_array **arrays,
- const struct _mesa_prim *prims,
- GLuint nr_prims,
- const struct _mesa_index_buffer *ib,
- GLboolean index_bounds_valid,
- GLuint min_index,
- GLuint max_index)
-{
- struct st_context *st = st_context(ctx);
- struct pipe_context *pipe = st->pipe;
- struct pipe_index_buffer ibuffer;
- struct pipe_draw_info info;
- unsigned i;
- GLboolean new_array = GL_TRUE;
- /* Fix this (Bug 34378):
- GLboolean new_array =
- st->dirty.st && (st->dirty.mesa & (_NEW_ARRAY | _NEW_PROGRAM)) != 0;*/
-
- /* Mesa core state should have been validated already */
- assert(ctx->NewState == 0x0);
-
- if (ib) {
- /* Gallium probably doesn't want this in some cases. */
- if (!index_bounds_valid)
- if (!vbo_all_varyings_in_vbos(arrays))
- vbo_get_minmax_index(ctx, prims, ib, &min_index, &max_index);
- } else {
- /* Get min/max index for non-indexed drawing. */
- min_index = ~0;
- max_index = 0;
-
- for (i = 0; i < nr_prims; i++) {
- min_index = MIN2(min_index, prims[i].start);
- max_index = MAX2(max_index, prims[i].start + prims[i].count - 1);
- }
- }
-
- /* Validate state. */
- if (st->dirty.st) {
- GLboolean vertDataEdgeFlags;
-
- /* sanity check for pointer arithmetic below */
- assert(sizeof(arrays[0]->Ptr[0]) == 1);
-
- vertDataEdgeFlags = arrays[VERT_ATTRIB_EDGEFLAG]->BufferObj &&
- arrays[VERT_ATTRIB_EDGEFLAG]->BufferObj->Name;
- if (vertDataEdgeFlags != st->vertdata_edgeflags) {
- st->vertdata_edgeflags = vertDataEdgeFlags;
- st->dirty.st |= ST_NEW_EDGEFLAGS_DATA;
- }
-
- st_validate_state(st);
-
- if (new_array) {
- st_validate_varrays(ctx, arrays, max_index);
- }
-
-#if 0
- if (MESA_VERBOSE & VERBOSE_GLSL) {
- check_uniforms(ctx);
- }
-#else
- (void) check_uniforms;
-#endif
- }
-
- /* Notify the driver that the content of user buffers may have been
- * changed. */
- if (!new_array && st->num_user_vbs) {
- for (i = 0; i < st->num_user_vbs; i++) {
- if (st->user_vb[i]) {
- unsigned stride = st->user_vb_stride[i];
-
- if (stride) {
- pipe->redefine_user_buffer(pipe, st->user_vb[i],
- min_index * stride,
- (max_index + 1 - min_index) * stride);
- } else {
- /* stride == 0 */
- pipe->redefine_user_buffer(pipe, st->user_vb[i],
- 0, st->user_vb[i]->width0);
- }
- }
- }
- }
-
- setup_index_buffer(ctx, ib, &ibuffer);
- pipe->set_index_buffer(pipe, &ibuffer);
-
- util_draw_init_info(&info);
- if (ib) {
- info.indexed = TRUE;
- if (min_index != ~0 && max_index != ~0) {
- info.min_index = min_index;
- info.max_index = max_index;
- }
- }
-
- info.primitive_restart = st->ctx->Array.PrimitiveRestart;
- info.restart_index = st->ctx->Array.RestartIndex;
-
- /* do actual drawing */
- for (i = 0; i < nr_prims; i++) {
- info.mode = translate_prim( ctx, prims[i].mode );
- info.start = prims[i].start;
- info.count = prims[i].count;
- info.instance_count = prims[i].num_instances;
- info.index_bias = prims[i].basevertex;
- if (!ib) {
- info.min_index = info.start;
- info.max_index = info.start + info.count - 1;
- }
-
- if (u_trim_pipe_prim(info.mode, &info.count))
- pipe->draw_vbo(pipe, &info);
- }
-
- pipe_resource_reference(&ibuffer.buffer, NULL);
-}
-
-
-void st_init_draw( struct st_context *st )
-{
- struct gl_context *ctx = st->ctx;
-
- vbo_set_draw_func(ctx, st_draw_vbo);
-
-#if FEATURE_feedback || FEATURE_rastpos
- st->draw = draw_create(st->pipe); /* for selection/feedback */
-
- /* Disable draw options that might convert points/lines to tris, etc.
- * as that would foul-up feedback/selection mode.
- */
- draw_wide_line_threshold(st->draw, 1000.0f);
- draw_wide_point_threshold(st->draw, 1000.0f);
- draw_enable_line_stipple(st->draw, FALSE);
- draw_enable_point_sprites(st->draw, FALSE);
-#endif
-}
-
-
-void st_destroy_draw( struct st_context *st )
-{
-#if FEATURE_feedback || FEATURE_rastpos
- draw_destroy(st->draw);
-#endif
-}
-
-
+/************************************************************************** + * + * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +/* + * This file implements the st_draw_vbo() function which is called from + * Mesa's VBO module. All point/line/triangle rendering is done through + * this function whether the user called glBegin/End, glDrawArrays, + * glDrawElements, glEvalMesh, or glCalList, etc. + * + * We basically convert the VBO's vertex attribute/array information into + * Gallium vertex state, bind the vertex buffer objects and call + * pipe->draw_elements(), pipe->draw_range_elements() or pipe->draw_arrays(). + * + * Authors: + * Keith Whitwell <keith@tungstengraphics.com> + */ + + +#include "main/imports.h" +#include "main/image.h" +#include "main/macros.h" +#include "main/mfeatures.h" +#include "program/prog_uniform.h" + +#include "vbo/vbo.h" + +#include "st_context.h" +#include "st_atom.h" +#include "st_cb_bufferobjects.h" +#include "st_draw.h" +#include "st_program.h" + +#include "pipe/p_context.h" +#include "pipe/p_defines.h" +#include "util/u_inlines.h" +#include "util/u_format.h" +#include "util/u_prim.h" +#include "util/u_draw_quad.h" +#include "draw/draw_context.h" +#include "cso_cache/cso_context.h" + + +static GLuint double_types[4] = { + PIPE_FORMAT_R64_FLOAT, + PIPE_FORMAT_R64G64_FLOAT, + PIPE_FORMAT_R64G64B64_FLOAT, + PIPE_FORMAT_R64G64B64A64_FLOAT +}; + +static GLuint float_types[4] = { + PIPE_FORMAT_R32_FLOAT, + PIPE_FORMAT_R32G32_FLOAT, + PIPE_FORMAT_R32G32B32_FLOAT, + PIPE_FORMAT_R32G32B32A32_FLOAT +}; + +static GLuint half_float_types[4] = { + PIPE_FORMAT_R16_FLOAT, + PIPE_FORMAT_R16G16_FLOAT, + PIPE_FORMAT_R16G16B16_FLOAT, + PIPE_FORMAT_R16G16B16A16_FLOAT +}; + +static GLuint uint_types_norm[4] = { + PIPE_FORMAT_R32_UNORM, + PIPE_FORMAT_R32G32_UNORM, + PIPE_FORMAT_R32G32B32_UNORM, + PIPE_FORMAT_R32G32B32A32_UNORM +}; + +static GLuint uint_types_scale[4] = { + PIPE_FORMAT_R32_USCALED, + PIPE_FORMAT_R32G32_USCALED, + PIPE_FORMAT_R32G32B32_USCALED, + PIPE_FORMAT_R32G32B32A32_USCALED +}; + +static GLuint int_types_norm[4] = { + PIPE_FORMAT_R32_SNORM, + PIPE_FORMAT_R32G32_SNORM, + PIPE_FORMAT_R32G32B32_SNORM, + PIPE_FORMAT_R32G32B32A32_SNORM +}; + +static GLuint int_types_scale[4] = { + PIPE_FORMAT_R32_SSCALED, + PIPE_FORMAT_R32G32_SSCALED, + PIPE_FORMAT_R32G32B32_SSCALED, + PIPE_FORMAT_R32G32B32A32_SSCALED +}; + +static GLuint ushort_types_norm[4] = { + PIPE_FORMAT_R16_UNORM, + PIPE_FORMAT_R16G16_UNORM, + PIPE_FORMAT_R16G16B16_UNORM, + PIPE_FORMAT_R16G16B16A16_UNORM +}; + +static GLuint ushort_types_scale[4] = { + PIPE_FORMAT_R16_USCALED, + PIPE_FORMAT_R16G16_USCALED, + PIPE_FORMAT_R16G16B16_USCALED, + PIPE_FORMAT_R16G16B16A16_USCALED +}; + +static GLuint short_types_norm[4] = { + PIPE_FORMAT_R16_SNORM, + PIPE_FORMAT_R16G16_SNORM, + PIPE_FORMAT_R16G16B16_SNORM, + PIPE_FORMAT_R16G16B16A16_SNORM +}; + +static GLuint short_types_scale[4] = { + PIPE_FORMAT_R16_SSCALED, + PIPE_FORMAT_R16G16_SSCALED, + PIPE_FORMAT_R16G16B16_SSCALED, + PIPE_FORMAT_R16G16B16A16_SSCALED +}; + +static GLuint ubyte_types_norm[4] = { + PIPE_FORMAT_R8_UNORM, + PIPE_FORMAT_R8G8_UNORM, + PIPE_FORMAT_R8G8B8_UNORM, + PIPE_FORMAT_R8G8B8A8_UNORM +}; + +static GLuint ubyte_types_scale[4] = { + PIPE_FORMAT_R8_USCALED, + PIPE_FORMAT_R8G8_USCALED, + PIPE_FORMAT_R8G8B8_USCALED, + PIPE_FORMAT_R8G8B8A8_USCALED +}; + +static GLuint byte_types_norm[4] = { + PIPE_FORMAT_R8_SNORM, + PIPE_FORMAT_R8G8_SNORM, + PIPE_FORMAT_R8G8B8_SNORM, + PIPE_FORMAT_R8G8B8A8_SNORM +}; + +static GLuint byte_types_scale[4] = { + PIPE_FORMAT_R8_SSCALED, + PIPE_FORMAT_R8G8_SSCALED, + PIPE_FORMAT_R8G8B8_SSCALED, + PIPE_FORMAT_R8G8B8A8_SSCALED +}; + +static GLuint fixed_types[4] = { + PIPE_FORMAT_R32_FIXED, + PIPE_FORMAT_R32G32_FIXED, + PIPE_FORMAT_R32G32B32_FIXED, + PIPE_FORMAT_R32G32B32A32_FIXED +}; + + + +/** + * Return a PIPE_FORMAT_x for the given GL datatype and size. + */ +GLuint +st_pipe_vertex_format(GLenum type, GLuint size, GLenum format, + GLboolean normalized) +{ + assert((type >= GL_BYTE && type <= GL_DOUBLE) || + type == GL_FIXED || type == GL_HALF_FLOAT); + assert(size >= 1); + assert(size <= 4); + assert(format == GL_RGBA || format == GL_BGRA); + + if (format == GL_BGRA) { + /* this is an odd-ball case */ + assert(type == GL_UNSIGNED_BYTE); + assert(normalized); + return PIPE_FORMAT_B8G8R8A8_UNORM; + } + + if (normalized) { + switch (type) { + case GL_DOUBLE: return double_types[size-1]; + case GL_FLOAT: return float_types[size-1]; + case GL_HALF_FLOAT: return half_float_types[size-1]; + case GL_INT: return int_types_norm[size-1]; + case GL_SHORT: return short_types_norm[size-1]; + case GL_BYTE: return byte_types_norm[size-1]; + case GL_UNSIGNED_INT: return uint_types_norm[size-1]; + case GL_UNSIGNED_SHORT: return ushort_types_norm[size-1]; + case GL_UNSIGNED_BYTE: return ubyte_types_norm[size-1]; + case GL_FIXED: return fixed_types[size-1]; + default: assert(0); return 0; + } + } + else { + switch (type) { + case GL_DOUBLE: return double_types[size-1]; + case GL_FLOAT: return float_types[size-1]; + case GL_HALF_FLOAT: return half_float_types[size-1]; + case GL_INT: return int_types_scale[size-1]; + case GL_SHORT: return short_types_scale[size-1]; + case GL_BYTE: return byte_types_scale[size-1]; + case GL_UNSIGNED_INT: return uint_types_scale[size-1]; + case GL_UNSIGNED_SHORT: return ushort_types_scale[size-1]; + case GL_UNSIGNED_BYTE: return ubyte_types_scale[size-1]; + case GL_FIXED: return fixed_types[size-1]; + default: assert(0); return 0; + } + } + return 0; /* silence compiler warning */ +} + + + + + +/** + * Examine the active arrays to determine if we have interleaved + * vertex arrays all living in one VBO, or all living in user space. + * \param userSpace returns whether the arrays are in user space. + */ +static GLboolean +is_interleaved_arrays(const struct st_vertex_program *vp, + const struct st_vp_variant *vpv, + const struct gl_client_array **arrays) +{ + GLuint attr; + const struct gl_buffer_object *firstBufObj = NULL; + GLint firstStride = -1; + const GLubyte *client_addr = NULL; + GLboolean user_memory; + + for (attr = 0; attr < vpv->num_inputs; attr++) { + const GLuint mesaAttr = vp->index_to_input[attr]; + const struct gl_buffer_object *bufObj = arrays[mesaAttr]->BufferObj; + const GLsizei stride = arrays[mesaAttr]->StrideB; /* in bytes */ + + if (firstStride < 0) { + firstStride = stride; + user_memory = !bufObj || !bufObj->Name; + } + else if (firstStride != stride) { + return GL_FALSE; + } + + if (!bufObj || !bufObj->Name) { + /* Try to detect if the client-space arrays are + * "close" to each other. + */ + if (!user_memory) { + return GL_FALSE; + } + if (!client_addr) { + client_addr = arrays[mesaAttr]->Ptr; + } + else if (abs(arrays[mesaAttr]->Ptr - client_addr) > firstStride) { + /* arrays start too far apart */ + return GL_FALSE; + } + } + else if (!firstBufObj) { + if (user_memory) { + return GL_FALSE; + } + firstBufObj = bufObj; + } + else if (bufObj != firstBufObj) { + return GL_FALSE; + } + } + + return GL_TRUE; +} + + +/** + * Set up for drawing interleaved arrays that all live in one VBO + * or all live in user space. + * \param vbuffer returns vertex buffer info + * \param velements returns vertex element info + */ +static void +setup_interleaved_attribs(struct gl_context *ctx, + const struct st_vertex_program *vp, + const struct st_vp_variant *vpv, + const struct gl_client_array **arrays, + struct pipe_vertex_buffer *vbuffer, + struct pipe_vertex_element velements[], + unsigned max_index) +{ + struct st_context *st = st_context(ctx); + struct pipe_context *pipe = st->pipe; + GLuint attr; + const GLubyte *low_addr = NULL; + + /* Find the lowest address. */ + for (attr = 0; attr < vpv->num_inputs; attr++) { + const GLubyte *start = arrays[vp->index_to_input[attr]]->Ptr; + + low_addr = !low_addr ? start : MIN2(low_addr, start); + } + + for (attr = 0; attr < vpv->num_inputs; attr++) { + const GLuint mesaAttr = vp->index_to_input[attr]; + struct gl_buffer_object *bufobj = arrays[mesaAttr]->BufferObj; + struct st_buffer_object *stobj = st_buffer_object(bufobj); + GLsizei stride = arrays[mesaAttr]->StrideB; + + if (attr == 0) { + if (bufobj && bufobj->Name) { + vbuffer->buffer = NULL; + pipe_resource_reference(&vbuffer->buffer, stobj->buffer); + vbuffer->buffer_offset = pointer_to_offset(low_addr); + } else { + vbuffer->buffer = + pipe_user_buffer_create(pipe->screen, (void*)low_addr, + stride * (max_index + 1), + PIPE_BIND_VERTEX_BUFFER); + vbuffer->buffer_offset = 0; + + /* Track user vertex buffers. */ + pipe_resource_reference(&st->user_vb[0], vbuffer->buffer); + st->user_vb_stride[0] = stride; + st->num_user_vbs = 1; + } + vbuffer->stride = stride; /* in bytes */ + } + + velements[attr].src_offset = + (unsigned) (arrays[mesaAttr]->Ptr - low_addr); + velements[attr].instance_divisor = arrays[mesaAttr]->InstanceDivisor; + velements[attr].vertex_buffer_index = 0; + velements[attr].src_format = + st_pipe_vertex_format(arrays[mesaAttr]->Type, + arrays[mesaAttr]->Size, + arrays[mesaAttr]->Format, + arrays[mesaAttr]->Normalized); + assert(velements[attr].src_format); + } +} + + +/** + * Set up a separate pipe_vertex_buffer and pipe_vertex_element for each + * vertex attribute. + * \param vbuffer returns vertex buffer info + * \param velements returns vertex element info + */ +static void +setup_non_interleaved_attribs(struct gl_context *ctx, + const struct st_vertex_program *vp, + const struct st_vp_variant *vpv, + const struct gl_client_array **arrays, + struct pipe_vertex_buffer vbuffer[], + struct pipe_vertex_element velements[], + unsigned max_index) +{ + struct st_context *st = st_context(ctx); + struct pipe_context *pipe = st->pipe; + GLuint attr; + + for (attr = 0; attr < vpv->num_inputs; attr++) { + const GLuint mesaAttr = vp->index_to_input[attr]; + struct gl_buffer_object *bufobj = arrays[mesaAttr]->BufferObj; + GLsizei stride = arrays[mesaAttr]->StrideB; + + if (bufobj && bufobj->Name) { + /* Attribute data is in a VBO. + * Recall that for VBOs, the gl_client_array->Ptr field is + * really an offset from the start of the VBO, not a pointer. + */ + struct st_buffer_object *stobj = st_buffer_object(bufobj); + assert(stobj->buffer); + + vbuffer[attr].buffer = NULL; + pipe_resource_reference(&vbuffer[attr].buffer, stobj->buffer); + vbuffer[attr].buffer_offset = pointer_to_offset(arrays[mesaAttr]->Ptr); + } + else { + /* wrap user data */ + if (arrays[mesaAttr]->Ptr) { + vbuffer[attr].buffer = + pipe_user_buffer_create(pipe->screen, + (void *) arrays[mesaAttr]->Ptr, + stride * (max_index + 1), + PIPE_BIND_VERTEX_BUFFER); + } + else { + /* no array, use ctx->Current.Attrib[] value */ + uint bytes = sizeof(ctx->Current.Attrib[0]); + vbuffer[attr].buffer = + pipe_user_buffer_create(pipe->screen, + (void *) ctx->Current.Attrib[mesaAttr], + bytes, + PIPE_BIND_VERTEX_BUFFER); + stride = 0; + } + + vbuffer[attr].buffer_offset = 0; + + /* Track user vertex buffers. */ + pipe_resource_reference(&st->user_vb[attr], vbuffer->buffer); + st->user_vb_stride[attr] = stride; + st->num_user_vbs = MAX2(st->num_user_vbs, attr+1); + } + + /* common-case setup */ + vbuffer[attr].stride = stride; /* in bytes */ + + velements[attr].src_offset = 0; + velements[attr].instance_divisor = arrays[mesaAttr]->InstanceDivisor; + velements[attr].vertex_buffer_index = attr; + velements[attr].src_format + = st_pipe_vertex_format(arrays[mesaAttr]->Type, + arrays[mesaAttr]->Size, + arrays[mesaAttr]->Format, + arrays[mesaAttr]->Normalized); + assert(velements[attr].src_format); + } +} + + +static void +setup_index_buffer(struct gl_context *ctx, + const struct _mesa_index_buffer *ib, + struct pipe_index_buffer *ibuffer) +{ + struct st_context *st = st_context(ctx); + struct pipe_context *pipe = st->pipe; + + memset(ibuffer, 0, sizeof(*ibuffer)); + if (ib) { + struct gl_buffer_object *bufobj = ib->obj; + + switch (ib->type) { + case GL_UNSIGNED_INT: + ibuffer->index_size = 4; + break; + case GL_UNSIGNED_SHORT: + ibuffer->index_size = 2; + break; + case GL_UNSIGNED_BYTE: + ibuffer->index_size = 1; + break; + default: + assert(0); + return; + } + + /* get/create the index buffer object */ + if (bufobj && bufobj->Name) { + /* elements/indexes are in a real VBO */ + struct st_buffer_object *stobj = st_buffer_object(bufobj); + pipe_resource_reference(&ibuffer->buffer, stobj->buffer); + ibuffer->offset = pointer_to_offset(ib->ptr); + } + else { + /* element/indicies are in user space memory */ + ibuffer->buffer = + pipe_user_buffer_create(pipe->screen, (void *) ib->ptr, + ib->count * ibuffer->index_size, + PIPE_BIND_INDEX_BUFFER); + } + } +} + +/** + * Prior to drawing, check that any uniforms referenced by the + * current shader have been set. If a uniform has not been set, + * issue a warning. + */ +static void +check_uniforms(struct gl_context *ctx) +{ + struct gl_shader_program *shProg[3] = { + ctx->Shader.CurrentVertexProgram, + ctx->Shader.CurrentGeometryProgram, + ctx->Shader.CurrentFragmentProgram, + }; + unsigned j; + + for (j = 0; j < 3; j++) { + unsigned i; + + if (shProg[j] == NULL || !shProg[j]->LinkStatus) + continue; + + for (i = 0; i < shProg[j]->Uniforms->NumUniforms; i++) { + const struct gl_uniform *u = &shProg[j]->Uniforms->Uniforms[i]; + if (!u->Initialized) { + _mesa_warning(ctx, + "Using shader with uninitialized uniform: %s", + u->Name); + } + } + } +} + + +/** + * Translate OpenGL primtive type (GL_POINTS, GL_TRIANGLE_STRIP, etc) to + * the corresponding Gallium type. + */ +static unsigned +translate_prim(const struct gl_context *ctx, unsigned prim) +{ + /* GL prims should match Gallium prims, spot-check a few */ + assert(GL_POINTS == PIPE_PRIM_POINTS); + assert(GL_QUADS == PIPE_PRIM_QUADS); + assert(GL_TRIANGLE_STRIP_ADJACENCY == PIPE_PRIM_TRIANGLE_STRIP_ADJACENCY); + + /* Avoid quadstrips if it's easy to do so: + * Note: it's imporant to do the correct trimming if we change the prim type! + * We do that wherever this function is called. + */ + if (prim == GL_QUAD_STRIP && + ctx->Light.ShadeModel != GL_FLAT && + ctx->Polygon.FrontMode == GL_FILL && + ctx->Polygon.BackMode == GL_FILL) + prim = GL_TRIANGLE_STRIP; + + return prim; +} + + +static void +st_validate_varrays(struct gl_context *ctx, + const struct gl_client_array **arrays, + unsigned max_index) +{ + struct st_context *st = st_context(ctx); + const struct st_vertex_program *vp; + const struct st_vp_variant *vpv; + struct pipe_vertex_buffer vbuffer[PIPE_MAX_SHADER_INPUTS]; + struct pipe_vertex_element velements[PIPE_MAX_ATTRIBS]; + unsigned num_vbuffers, num_velements; + GLuint attr; + unsigned i; + + /* must get these after state validation! */ + vp = st->vp; + vpv = st->vp_variant; + + memset(velements, 0, sizeof(struct pipe_vertex_element) * vpv->num_inputs); + + /* Unreference any user vertex buffers. */ + for (i = 0; i < st->num_user_vbs; i++) { + pipe_resource_reference(&st->user_vb[i], NULL); + } + st->num_user_vbs = 0; + + /* + * Setup the vbuffer[] and velements[] arrays. + */ + if (is_interleaved_arrays(vp, vpv, arrays)) { + setup_interleaved_attribs(ctx, vp, vpv, arrays, vbuffer, velements, + max_index); + num_vbuffers = 1; + num_velements = vpv->num_inputs; + if (num_velements == 0) + num_vbuffers = 0; + } + else { + setup_non_interleaved_attribs(ctx, vp, vpv, arrays, + vbuffer, velements, max_index); + num_vbuffers = vpv->num_inputs; + num_velements = vpv->num_inputs; + } + + cso_set_vertex_buffers(st->cso_context, num_vbuffers, vbuffer); + cso_set_vertex_elements(st->cso_context, num_velements, velements); + + /* unreference buffers (frees wrapped user-space buffer objects) + * This is OK, because the pipe driver should reference buffers by itself + * in set_vertex_buffers. */ + for (attr = 0; attr < num_vbuffers; attr++) { + pipe_resource_reference(&vbuffer[attr].buffer, NULL); + assert(!vbuffer[attr].buffer); + } +} + + +/** + * This function gets plugged into the VBO module and is called when + * we have something to render. + * Basically, translate the information into the format expected by gallium. + */ +void +st_draw_vbo(struct gl_context *ctx, + const struct gl_client_array **arrays, + const struct _mesa_prim *prims, + GLuint nr_prims, + const struct _mesa_index_buffer *ib, + GLboolean index_bounds_valid, + GLuint min_index, + GLuint max_index) +{ + struct st_context *st = st_context(ctx); + struct pipe_context *pipe = st->pipe; + struct pipe_index_buffer ibuffer; + struct pipe_draw_info info; + unsigned i; + GLboolean new_array = GL_TRUE; + /* Fix this (Bug 34378): + GLboolean new_array = + st->dirty.st && (st->dirty.mesa & (_NEW_ARRAY | _NEW_PROGRAM)) != 0;*/ + + /* Mesa core state should have been validated already */ + assert(ctx->NewState == 0x0); + + if (ib) { + /* Gallium probably doesn't want this in some cases. */ + if (!index_bounds_valid) + if (!vbo_all_varyings_in_vbos(arrays)) + vbo_get_minmax_index(ctx, prims, ib, &min_index, &max_index); + } else { + /* Get min/max index for non-indexed drawing. */ + min_index = ~0; + max_index = 0; + + for (i = 0; i < nr_prims; i++) { + min_index = MIN2(min_index, prims[i].start); + max_index = MAX2(max_index, prims[i].start + prims[i].count - 1); + } + } + + /* Validate state. */ + if (st->dirty.st) { + GLboolean vertDataEdgeFlags; + + /* sanity check for pointer arithmetic below */ + assert(sizeof(arrays[0]->Ptr[0]) == 1); + + vertDataEdgeFlags = arrays[VERT_ATTRIB_EDGEFLAG]->BufferObj && + arrays[VERT_ATTRIB_EDGEFLAG]->BufferObj->Name; + if (vertDataEdgeFlags != st->vertdata_edgeflags) { + st->vertdata_edgeflags = vertDataEdgeFlags; + st->dirty.st |= ST_NEW_EDGEFLAGS_DATA; + } + + st_validate_state(st); + + if (new_array) { + st_validate_varrays(ctx, arrays, max_index); + } + +#if 0 + if (MESA_VERBOSE & VERBOSE_GLSL) { + check_uniforms(ctx); + } +#else + (void) check_uniforms; +#endif + } + + /* Notify the driver that the content of user buffers may have been + * changed. */ + if (!new_array && st->num_user_vbs) { + for (i = 0; i < st->num_user_vbs; i++) { + if (st->user_vb[i]) { + unsigned stride = st->user_vb_stride[i]; + + if (stride) { + pipe->redefine_user_buffer(pipe, st->user_vb[i], + min_index * stride, + (max_index + 1 - min_index) * stride); + } else { + /* stride == 0 */ + pipe->redefine_user_buffer(pipe, st->user_vb[i], + 0, st->user_vb[i]->width0); + } + } + } + } + + setup_index_buffer(ctx, ib, &ibuffer); + pipe->set_index_buffer(pipe, &ibuffer); + + util_draw_init_info(&info); + if (ib) { + info.indexed = TRUE; + if (min_index != ~0 && max_index != ~0) { + info.min_index = min_index; + info.max_index = max_index; + } + } + + info.primitive_restart = st->ctx->Array.PrimitiveRestart; + info.restart_index = st->ctx->Array.RestartIndex; + + /* do actual drawing */ + for (i = 0; i < nr_prims; i++) { + info.mode = translate_prim( ctx, prims[i].mode ); + info.start = prims[i].start; + info.count = prims[i].count; + info.instance_count = prims[i].num_instances; + info.index_bias = prims[i].basevertex; + if (!ib) { + info.min_index = info.start; + info.max_index = info.start + info.count - 1; + } + + if (u_trim_pipe_prim(info.mode, &info.count)) + pipe->draw_vbo(pipe, &info); + } + + pipe_resource_reference(&ibuffer.buffer, NULL); +} + + +void st_init_draw( struct st_context *st ) +{ + struct gl_context *ctx = st->ctx; + + vbo_set_draw_func(ctx, st_draw_vbo); + +#if FEATURE_feedback || FEATURE_rastpos + st->draw = draw_create(st->pipe); /* for selection/feedback */ + + /* Disable draw options that might convert points/lines to tris, etc. + * as that would foul-up feedback/selection mode. + */ + draw_wide_line_threshold(st->draw, 1000.0f); + draw_wide_point_threshold(st->draw, 1000.0f); + draw_enable_line_stipple(st->draw, FALSE); + draw_enable_point_sprites(st->draw, FALSE); +#endif +} + + +void st_destroy_draw( struct st_context *st ) +{ +#if FEATURE_feedback || FEATURE_rastpos + draw_destroy(st->draw); +#endif +} + + diff --git a/mesalib/src/mesa/state_tracker/st_format.c b/mesalib/src/mesa/state_tracker/st_format.c index a532e089c..577ee6189 100644 --- a/mesalib/src/mesa/state_tracker/st_format.c +++ b/mesalib/src/mesa/state_tracker/st_format.c @@ -113,12 +113,9 @@ st_format_datatype(enum pipe_format format) return GL_UNSIGNED_SHORT; } else { - /* compressed format? */ - assert(0); + /* probably a compressed format, unsupported anyway */ + return GL_NONE; } - - assert(0); - return GL_NONE; } @@ -1098,3 +1095,55 @@ st_sampler_compat_formats(enum pipe_format format1, enum pipe_format format2) return GL_FALSE; } + + + +/** + * This is used for translating texture border color and the clear + * color. For example, the clear color is interpreted according to + * the renderbuffer's base format. For example, if clearing a + * GL_LUMINANCE buffer, ClearColor[0] = luminance and ClearColor[1] = + * alpha. Similarly for texture border colors. + */ +void +st_translate_color(const GLfloat colorIn[4], GLenum baseFormat, + GLfloat colorOut[4]) +{ + switch (baseFormat) { + case GL_RED: + colorOut[0] = colorIn[0]; + colorOut[1] = 0.0F; + colorOut[2] = 0.0F; + colorOut[3] = 1.0F; + break; + case GL_RG: + colorOut[0] = colorIn[0]; + colorOut[1] = colorIn[1]; + colorOut[2] = 0.0F; + colorOut[3] = 1.0F; + break; + case GL_RGB: + colorOut[0] = colorIn[0]; + colorOut[1] = colorIn[1]; + colorOut[2] = colorIn[2]; + colorOut[3] = 1.0F; + break; + case GL_ALPHA: + colorOut[0] = colorOut[1] = colorOut[2] = 0.0; + colorOut[3] = colorIn[3]; + break; + case GL_LUMINANCE: + colorOut[0] = colorOut[1] = colorOut[2] = colorIn[0]; + colorOut[3] = 1.0; + break; + case GL_LUMINANCE_ALPHA: + colorOut[0] = colorOut[1] = colorOut[2] = colorIn[0]; + colorOut[3] = colorIn[3]; + break; + case GL_INTENSITY: + colorOut[0] = colorOut[1] = colorOut[2] = colorOut[3] = colorIn[0]; + break; + default: + COPY_4V(colorOut, colorIn); + } +} diff --git a/mesalib/src/mesa/state_tracker/st_format.h b/mesalib/src/mesa/state_tracker/st_format.h index ffcaf402a..0fb570f6e 100644 --- a/mesalib/src/mesa/state_tracker/st_format.h +++ b/mesalib/src/mesa/state_tracker/st_format.h @@ -1,80 +1,86 @@ -/**************************************************************************
- *
- * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
- * Copyright (c) 2010 VMware, Inc.
- * All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- **************************************************************************/
-
-
-#ifndef ST_FORMAT_H
-#define ST_FORMAT_H
-
-#include "main/formats.h"
-#include "main/glheader.h"
-
-#include "pipe/p_defines.h"
-#include "pipe/p_format.h"
-
-struct gl_context;
-struct pipe_screen;
-
-extern GLenum
-st_format_datatype(enum pipe_format format);
-
-
-extern enum pipe_format
-st_mesa_format_to_pipe_format(gl_format mesaFormat);
-
-extern gl_format
-st_pipe_format_to_mesa_format(enum pipe_format pipeFormat);
-
-
-extern enum pipe_format
-st_choose_format(struct pipe_screen *screen, GLenum internalFormat,
- enum pipe_texture_target target, unsigned sample_count,
- unsigned tex_usage);
-
-extern enum pipe_format
-st_choose_renderbuffer_format(struct pipe_screen *screen,
- GLenum internalFormat, unsigned sample_count);
-
-
-gl_format
-st_ChooseTextureFormat_renderable(struct gl_context *ctx, GLint internalFormat,
- GLenum format, GLenum type, GLboolean renderable);
-
-extern gl_format
-st_ChooseTextureFormat(struct gl_context * ctx, GLint internalFormat,
- GLenum format, GLenum type);
-
-
-extern GLboolean
-st_equal_formats(enum pipe_format pFormat, GLenum format, GLenum type);
-
-/* can we use a sampler view to translate these formats
- only used to make TFP so far */
-extern GLboolean
-st_sampler_compat_formats(enum pipe_format format1, enum pipe_format format2);
-
-#endif /* ST_FORMAT_H */
+/************************************************************************** + * + * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. + * Copyright (c) 2010 VMware, Inc. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + + +#ifndef ST_FORMAT_H +#define ST_FORMAT_H + +#include "main/formats.h" +#include "main/glheader.h" + +#include "pipe/p_defines.h" +#include "pipe/p_format.h" + +struct gl_context; +struct pipe_screen; + +extern GLenum +st_format_datatype(enum pipe_format format); + + +extern enum pipe_format +st_mesa_format_to_pipe_format(gl_format mesaFormat); + +extern gl_format +st_pipe_format_to_mesa_format(enum pipe_format pipeFormat); + + +extern enum pipe_format +st_choose_format(struct pipe_screen *screen, GLenum internalFormat, + enum pipe_texture_target target, unsigned sample_count, + unsigned tex_usage); + +extern enum pipe_format +st_choose_renderbuffer_format(struct pipe_screen *screen, + GLenum internalFormat, unsigned sample_count); + + +gl_format +st_ChooseTextureFormat_renderable(struct gl_context *ctx, GLint internalFormat, + GLenum format, GLenum type, GLboolean renderable); + +extern gl_format +st_ChooseTextureFormat(struct gl_context * ctx, GLint internalFormat, + GLenum format, GLenum type); + + +extern GLboolean +st_equal_formats(enum pipe_format pFormat, GLenum format, GLenum type); + +/* can we use a sampler view to translate these formats + only used to make TFP so far */ +extern GLboolean +st_sampler_compat_formats(enum pipe_format format1, enum pipe_format format2); + + +extern void +st_translate_color(const GLfloat colorIn[4], GLenum baseFormat, + GLfloat colorOut[4]); + + +#endif /* ST_FORMAT_H */ diff --git a/mesalib/src/mesa/state_tracker/st_gen_mipmap.c b/mesalib/src/mesa/state_tracker/st_gen_mipmap.c index 18eb3be68..4bf682808 100644 --- a/mesalib/src/mesa/state_tracker/st_gen_mipmap.c +++ b/mesalib/src/mesa/state_tracker/st_gen_mipmap.c @@ -336,6 +336,11 @@ st_generate_mipmap(struct gl_context *ctx, GLenum target, if (lastLevel == 0) return; + /* The texture isn't in a "complete" state yet so set the expected + * lastLevel here, since it won't get done in st_finalize_texture(). + */ + stObj->lastLevel = lastLevel; + if (pt->last_level < lastLevel) { /* The current gallium texture doesn't have space for all the * mipmap levels we need to generate. So allocate a new texture. @@ -353,11 +358,6 @@ st_generate_mipmap(struct gl_context *ctx, GLenum target, oldTex->array_size, oldTex->bind); - /* The texture isn't in a "complete" state yet so set the expected - * lastLevel here, since it won't get done in st_finalize_texture(). - */ - stObj->lastLevel = lastLevel; - /* This will copy the old texture's base image into the new texture * which we just allocated. */ @@ -366,8 +366,6 @@ st_generate_mipmap(struct gl_context *ctx, GLenum target, /* release the old tex (will likely be freed too) */ pipe_resource_reference(&oldTex, NULL); pipe_sampler_view_reference(&stObj->sampler_view, NULL); - - pt = stObj->pt; } else { /* Make sure that the base texture image data is present in the @@ -376,6 +374,8 @@ st_generate_mipmap(struct gl_context *ctx, GLenum target, st_finalize_texture(ctx, st->pipe, texObj); } + pt = stObj->pt; + assert(pt->last_level >= lastLevel); /* Try to generate the mipmap by rendering/texturing. If that fails, diff --git a/mesalib/src/mesa/vbo/vbo_exec.h b/mesalib/src/mesa/vbo/vbo_exec.h index 2508016c0..ca100e428 100644 --- a/mesalib/src/mesa/vbo/vbo_exec.h +++ b/mesalib/src/mesa/vbo/vbo_exec.h @@ -1,199 +1,198 @@ -/**************************************************************************
-
-Copyright 2002 Tungsten Graphics Inc., Cedar Park, Texas.
-
-All Rights Reserved.
-
-Permission is hereby granted, free of charge, to any person obtaining a
-copy of this software and associated documentation files (the "Software"),
-to deal in the Software without restriction, including without limitation
-on the rights to use, copy, modify, merge, publish, distribute, sub
-license, and/or sell copies of the Software, and to permit persons to whom
-the Software is furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice (including the next
-paragraph) shall be included in all copies or substantial portions of the
-Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
-TUNGSTEN GRAPHICS AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM,
-DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
-OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
-USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-/*
- * Authors:
- * Keith Whitwell <keith@tungstengraphics.com>
- *
- */
-
-#ifndef __VBO_EXEC_H__
-#define __VBO_EXEC_H__
-
-#include "main/mfeatures.h"
-#include "main/mtypes.h"
-#include "vbo.h"
-#include "vbo_attrib.h"
-
-
-#define VBO_MAX_PRIM 64
-
-/* Wierd implementation stuff:
- */
-#define VBO_VERT_BUFFER_SIZE (1024*64) /* bytes */
-#define VBO_MAX_ATTR_CODEGEN 16
-#define ERROR_ATTRIB 16
-
-
-/** Current vertex program mode */
-enum vp_mode {
- VP_NONE, /**< fixed function */
- VP_NV, /**< NV vertex program */
- VP_ARB /**< ARB vertex program or GLSL vertex shader */
-};
-
-
-struct vbo_exec_eval1_map {
- struct gl_1d_map *map;
- GLuint sz;
-};
-
-struct vbo_exec_eval2_map {
- struct gl_2d_map *map;
- GLuint sz;
-};
-
-
-
-struct vbo_exec_copied_vtx {
- GLfloat buffer[VBO_ATTRIB_MAX * 4 * VBO_MAX_COPIED_VERTS];
- GLuint nr;
-};
-
-
-typedef void (*vbo_attrfv_func)( const GLfloat * );
-
-
-struct vbo_exec_context
-{
- struct gl_context *ctx;
- GLvertexformat vtxfmt;
-
- struct {
- struct gl_buffer_object *bufferobj;
-
- GLuint vertex_size; /* in dwords */
-
- struct _mesa_prim prim[VBO_MAX_PRIM];
- GLuint prim_count;
-
- GLfloat *buffer_map;
- GLfloat *buffer_ptr; /* cursor, points into buffer */
- GLuint buffer_used; /* in bytes */
- GLfloat vertex[VBO_ATTRIB_MAX*4]; /* current vertex */
-
- GLuint vert_count;
- GLuint max_vert;
- struct vbo_exec_copied_vtx copied;
-
- GLubyte attrsz[VBO_ATTRIB_MAX];
- GLubyte active_sz[VBO_ATTRIB_MAX];
-
- GLfloat *attrptr[VBO_ATTRIB_MAX];
- struct gl_client_array arrays[VERT_ATTRIB_MAX];
-
- /* According to program mode, the values above plus current
- * values are squashed down to the 32 attributes passed to the
- * vertex program below:
- */
- enum vp_mode program_mode;
- GLuint enabled_flags;
- const struct gl_client_array *inputs[VERT_ATTRIB_MAX];
- } vtx;
-
-
- struct {
- GLboolean recalculate_maps;
- struct vbo_exec_eval1_map map1[VERT_ATTRIB_MAX];
- struct vbo_exec_eval2_map map2[VERT_ATTRIB_MAX];
- } eval;
-
- struct {
- enum vp_mode program_mode;
- GLuint enabled_flags;
- GLuint array_obj;
-
- /* These just mirror the current arrayobj (todo: make arrayobj
- * look like this and remove the mirror):
- */
- const struct gl_client_array *legacy_array[16];
- const struct gl_client_array *generic_array[16];
-
- /* Arrays and current values manipulated according to program
- * mode, etc. These are the attributes as seen by vertex
- * programs:
- */
- const struct gl_client_array *inputs[VERT_ATTRIB_MAX];
- } array;
-
-#ifdef DEBUG
- GLint flush_call_depth;
-#endif
-};
-
-
-
-/* External API:
- */
-void vbo_exec_init( struct gl_context *ctx );
-void vbo_exec_destroy( struct gl_context *ctx );
-void vbo_exec_invalidate_state( struct gl_context *ctx, GLuint new_state );
-void vbo_exec_FlushVertices_internal( struct gl_context *ctx, GLboolean unmap );
-
-void vbo_exec_BeginVertices( struct gl_context *ctx );
-void vbo_exec_FlushVertices( struct gl_context *ctx, GLuint flags );
-
-
-/* Internal functions:
- */
-void vbo_exec_array_init( struct vbo_exec_context *exec );
-void vbo_exec_array_destroy( struct vbo_exec_context *exec );
-
-
-void vbo_exec_vtx_init( struct vbo_exec_context *exec );
-void vbo_exec_vtx_destroy( struct vbo_exec_context *exec );
-
-#if FEATURE_beginend
-
-void vbo_exec_vtx_flush( struct vbo_exec_context *exec, GLboolean unmap );
-void vbo_exec_vtx_map( struct vbo_exec_context *exec );
-
-#else /* FEATURE_beginend */
-
-static INLINE void
-vbo_exec_vtx_flush( struct vbo_exec_context *exec, GLboolean unmap )
-{
-}
-
-static INLINE void
-vbo_exec_vtx_map( struct vbo_exec_context *exec )
-{
-}
-
-#endif /* FEATURE_beginend */
-
-void vbo_exec_vtx_wrap( struct vbo_exec_context *exec );
-
-void vbo_exec_eval_update( struct vbo_exec_context *exec );
-
-void vbo_exec_do_EvalCoord2f( struct vbo_exec_context *exec,
- GLfloat u, GLfloat v );
-
-void vbo_exec_do_EvalCoord1f( struct vbo_exec_context *exec,
- GLfloat u);
-
-#endif
+/************************************************************************** + +Copyright 2002 Tungsten Graphics Inc., Cedar Park, Texas. + +All Rights Reserved. + +Permission is hereby granted, free of charge, to any person obtaining a +copy of this software and associated documentation files (the "Software"), +to deal in the Software without restriction, including without limitation +on the rights to use, copy, modify, merge, publish, distribute, sub +license, and/or sell copies of the Software, and to permit persons to whom +the Software is furnished to do so, subject to the following conditions: + +The above copyright notice and this permission notice (including the next +paragraph) shall be included in all copies or substantial portions of the +Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL +TUNGSTEN GRAPHICS AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM, +DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR +OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE +USE OR OTHER DEALINGS IN THE SOFTWARE. + +**************************************************************************/ + +/* + * Authors: + * Keith Whitwell <keith@tungstengraphics.com> + * + */ + +#ifndef __VBO_EXEC_H__ +#define __VBO_EXEC_H__ + +#include "main/mfeatures.h" +#include "main/mtypes.h" +#include "vbo.h" +#include "vbo_attrib.h" + + +#define VBO_MAX_PRIM 64 + +/* Wierd implementation stuff: + */ +#define VBO_VERT_BUFFER_SIZE (1024*64) /* bytes */ +#define VBO_MAX_ATTR_CODEGEN 16 +#define ERROR_ATTRIB 16 + + +/** Current vertex program mode */ +enum vp_mode { + VP_NONE, /**< fixed function */ + VP_NV, /**< NV vertex program */ + VP_ARB /**< ARB vertex program or GLSL vertex shader */ +}; + + +struct vbo_exec_eval1_map { + struct gl_1d_map *map; + GLuint sz; +}; + +struct vbo_exec_eval2_map { + struct gl_2d_map *map; + GLuint sz; +}; + + + +struct vbo_exec_copied_vtx { + GLfloat buffer[VBO_ATTRIB_MAX * 4 * VBO_MAX_COPIED_VERTS]; + GLuint nr; +}; + + +typedef void (*vbo_attrfv_func)( const GLfloat * ); + + +struct vbo_exec_context +{ + struct gl_context *ctx; + GLvertexformat vtxfmt; + + struct { + struct gl_buffer_object *bufferobj; + + GLuint vertex_size; /* in dwords */ + + struct _mesa_prim prim[VBO_MAX_PRIM]; + GLuint prim_count; + + GLfloat *buffer_map; + GLfloat *buffer_ptr; /* cursor, points into buffer */ + GLuint buffer_used; /* in bytes */ + GLfloat vertex[VBO_ATTRIB_MAX*4]; /* current vertex */ + + GLuint vert_count; + GLuint max_vert; + struct vbo_exec_copied_vtx copied; + + GLubyte attrsz[VBO_ATTRIB_MAX]; + GLubyte active_sz[VBO_ATTRIB_MAX]; + + GLfloat *attrptr[VBO_ATTRIB_MAX]; + struct gl_client_array arrays[VERT_ATTRIB_MAX]; + + /* According to program mode, the values above plus current + * values are squashed down to the 32 attributes passed to the + * vertex program below: + */ + enum vp_mode program_mode; + GLuint enabled_flags; + const struct gl_client_array *inputs[VERT_ATTRIB_MAX]; + } vtx; + + + struct { + GLboolean recalculate_maps; + struct vbo_exec_eval1_map map1[VERT_ATTRIB_MAX]; + struct vbo_exec_eval2_map map2[VERT_ATTRIB_MAX]; + } eval; + + struct { + enum vp_mode program_mode; + GLuint enabled_flags; + GLuint array_obj; + + /* These just mirror the current arrayobj (todo: make arrayobj + * look like this and remove the mirror): + */ + const struct gl_client_array *legacy_array[16]; + const struct gl_client_array *generic_array[16]; + + /* Arrays and current values manipulated according to program + * mode, etc. These are the attributes as seen by vertex + * programs: + */ + const struct gl_client_array *inputs[VERT_ATTRIB_MAX]; + } array; + +#ifdef DEBUG + GLint flush_call_depth; +#endif +}; + + + +/* External API: + */ +void vbo_exec_init( struct gl_context *ctx ); +void vbo_exec_destroy( struct gl_context *ctx ); +void vbo_exec_invalidate_state( struct gl_context *ctx, GLuint new_state ); + +void vbo_exec_BeginVertices( struct gl_context *ctx ); +void vbo_exec_FlushVertices( struct gl_context *ctx, GLuint flags ); + + +/* Internal functions: + */ +void vbo_exec_array_init( struct vbo_exec_context *exec ); +void vbo_exec_array_destroy( struct vbo_exec_context *exec ); + + +void vbo_exec_vtx_init( struct vbo_exec_context *exec ); +void vbo_exec_vtx_destroy( struct vbo_exec_context *exec ); + +#if FEATURE_beginend + +void vbo_exec_vtx_flush( struct vbo_exec_context *exec, GLboolean unmap ); +void vbo_exec_vtx_map( struct vbo_exec_context *exec ); + +#else /* FEATURE_beginend */ + +static INLINE void +vbo_exec_vtx_flush( struct vbo_exec_context *exec, GLboolean unmap ) +{ +} + +static INLINE void +vbo_exec_vtx_map( struct vbo_exec_context *exec ) +{ +} + +#endif /* FEATURE_beginend */ + +void vbo_exec_vtx_wrap( struct vbo_exec_context *exec ); + +void vbo_exec_eval_update( struct vbo_exec_context *exec ); + +void vbo_exec_do_EvalCoord2f( struct vbo_exec_context *exec, + GLfloat u, GLfloat v ); + +void vbo_exec_do_EvalCoord1f( struct vbo_exec_context *exec, + GLfloat u); + +#endif diff --git a/mesalib/src/mesa/vbo/vbo_exec_api.c b/mesalib/src/mesa/vbo/vbo_exec_api.c index 0c59c41ba..c4d39d8f1 100644 --- a/mesalib/src/mesa/vbo/vbo_exec_api.c +++ b/mesalib/src/mesa/vbo/vbo_exec_api.c @@ -1,1080 +1,1107 @@ -/**************************************************************************
-
-Copyright 2002-2008 Tungsten Graphics Inc., Cedar Park, Texas.
-
-All Rights Reserved.
-
-Permission is hereby granted, free of charge, to any person obtaining a
-copy of this software and associated documentation files (the "Software"),
-to deal in the Software without restriction, including without limitation
-on the rights to use, copy, modify, merge, publish, distribute, sub
-license, and/or sell copies of the Software, and to permit persons to whom
-the Software is furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice (including the next
-paragraph) shall be included in all copies or substantial portions of the
-Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
-TUNGSTEN GRAPHICS AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM,
-DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
-OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
-USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-/*
- * Authors:
- * Keith Whitwell <keith@tungstengraphics.com>
- */
-
-#include "main/glheader.h"
-#include "main/bufferobj.h"
-#include "main/context.h"
-#include "main/macros.h"
-#include "main/mfeatures.h"
-#include "main/vtxfmt.h"
-#include "main/dlist.h"
-#include "main/eval.h"
-#include "main/state.h"
-#include "main/light.h"
-#include "main/api_arrayelt.h"
-#include "main/api_noop.h"
-#include "main/dispatch.h"
-
-#include "vbo_context.h"
-
-#ifdef ERROR
-#undef ERROR
-#endif
-
-
-/** ID/name for immediate-mode VBO */
-#define IMM_BUFFER_NAME 0xaabbccdd
-
-
-static void reset_attrfv( struct vbo_exec_context *exec );
-
-
-/**
- * Close off the last primitive, execute the buffer, restart the
- * primitive.
- */
-static void vbo_exec_wrap_buffers( struct vbo_exec_context *exec )
-{
- if (exec->vtx.prim_count == 0) {
- exec->vtx.copied.nr = 0;
- exec->vtx.vert_count = 0;
- exec->vtx.buffer_ptr = exec->vtx.buffer_map;
- }
- else {
- GLuint last_begin = exec->vtx.prim[exec->vtx.prim_count-1].begin;
- GLuint last_count;
-
- if (exec->ctx->Driver.CurrentExecPrimitive != PRIM_OUTSIDE_BEGIN_END) {
- GLint i = exec->vtx.prim_count - 1;
- assert(i >= 0);
- exec->vtx.prim[i].count = (exec->vtx.vert_count -
- exec->vtx.prim[i].start);
- }
-
- last_count = exec->vtx.prim[exec->vtx.prim_count-1].count;
-
- /* Execute the buffer and save copied vertices.
- */
- if (exec->vtx.vert_count)
- vbo_exec_vtx_flush( exec, GL_FALSE );
- else {
- exec->vtx.prim_count = 0;
- exec->vtx.copied.nr = 0;
- }
-
- /* Emit a glBegin to start the new list.
- */
- assert(exec->vtx.prim_count == 0);
-
- if (exec->ctx->Driver.CurrentExecPrimitive != PRIM_OUTSIDE_BEGIN_END) {
- exec->vtx.prim[0].mode = exec->ctx->Driver.CurrentExecPrimitive;
- exec->vtx.prim[0].start = 0;
- exec->vtx.prim[0].count = 0;
- exec->vtx.prim_count++;
-
- if (exec->vtx.copied.nr == last_count)
- exec->vtx.prim[0].begin = last_begin;
- }
- }
-}
-
-
-/**
- * Deal with buffer wrapping where provoked by the vertex buffer
- * filling up, as opposed to upgrade_vertex().
- */
-void vbo_exec_vtx_wrap( struct vbo_exec_context *exec )
-{
- GLfloat *data = exec->vtx.copied.buffer;
- GLuint i;
-
- /* Run pipeline on current vertices, copy wrapped vertices
- * to exec->vtx.copied.
- */
- vbo_exec_wrap_buffers( exec );
-
- /* Copy stored stored vertices to start of new list.
- */
- assert(exec->vtx.max_vert - exec->vtx.vert_count > exec->vtx.copied.nr);
-
- for (i = 0 ; i < exec->vtx.copied.nr ; i++) {
- memcpy( exec->vtx.buffer_ptr, data,
- exec->vtx.vertex_size * sizeof(GLfloat));
- exec->vtx.buffer_ptr += exec->vtx.vertex_size;
- data += exec->vtx.vertex_size;
- exec->vtx.vert_count++;
- }
-
- exec->vtx.copied.nr = 0;
-}
-
-
-/**
- * Copy the active vertex's values to the ctx->Current fields.
- */
-static void vbo_exec_copy_to_current( struct vbo_exec_context *exec )
-{
- struct gl_context *ctx = exec->ctx;
- struct vbo_context *vbo = vbo_context(ctx);
- GLuint i;
-
- for (i = VBO_ATTRIB_POS+1 ; i < VBO_ATTRIB_MAX ; i++) {
- if (exec->vtx.attrsz[i]) {
- /* Note: the exec->vtx.current[i] pointers point into the
- * ctx->Current.Attrib and ctx->Light.Material.Attrib arrays.
- */
- GLfloat *current = (GLfloat *)vbo->currval[i].Ptr;
- GLfloat tmp[4];
-
- COPY_CLEAN_4V(tmp,
- exec->vtx.attrsz[i],
- exec->vtx.attrptr[i]);
-
- if (memcmp(current, tmp, sizeof(tmp)) != 0)
- {
- memcpy(current, tmp, sizeof(tmp));
-
- /* Given that we explicitly state size here, there is no need
- * for the COPY_CLEAN above, could just copy 16 bytes and be
- * done. The only problem is when Mesa accesses ctx->Current
- * directly.
- */
- vbo->currval[i].Size = exec->vtx.attrsz[i];
-
- /* This triggers rather too much recalculation of Mesa state
- * that doesn't get used (eg light positions).
- */
- if (i >= VBO_ATTRIB_MAT_FRONT_AMBIENT &&
- i <= VBO_ATTRIB_MAT_BACK_INDEXES)
- ctx->NewState |= _NEW_LIGHT;
-
- ctx->NewState |= _NEW_CURRENT_ATTRIB;
- }
- }
- }
-
- /* Colormaterial -- this kindof sucks.
- */
- if (ctx->Light.ColorMaterialEnabled &&
- exec->vtx.attrsz[VBO_ATTRIB_COLOR0]) {
- _mesa_update_color_material(ctx,
- ctx->Current.Attrib[VBO_ATTRIB_COLOR0]);
- }
-}
-
-
-static void vbo_exec_copy_from_current( struct vbo_exec_context *exec )
-{
- struct gl_context *ctx = exec->ctx;
- struct vbo_context *vbo = vbo_context(ctx);
- GLint i;
-
- for (i = VBO_ATTRIB_POS+1 ; i < VBO_ATTRIB_MAX ; i++) {
- const GLfloat *current = (GLfloat *)vbo->currval[i].Ptr;
- switch (exec->vtx.attrsz[i]) {
- case 4: exec->vtx.attrptr[i][3] = current[3];
- case 3: exec->vtx.attrptr[i][2] = current[2];
- case 2: exec->vtx.attrptr[i][1] = current[1];
- case 1: exec->vtx.attrptr[i][0] = current[0];
- break;
- }
- }
-}
-
-
-/**
- * Flush existing data, set new attrib size, replay copied vertices.
- */
-static void vbo_exec_wrap_upgrade_vertex( struct vbo_exec_context *exec,
- GLuint attr,
- GLuint newsz )
-{
- struct gl_context *ctx = exec->ctx;
- struct vbo_context *vbo = vbo_context(ctx);
- GLint lastcount = exec->vtx.vert_count;
- GLfloat *old_attrptr[VBO_ATTRIB_MAX];
- GLuint old_vtx_size = exec->vtx.vertex_size;
- GLuint oldsz = exec->vtx.attrsz[attr];
- GLuint i;
-
- /* Run pipeline on current vertices, copy wrapped vertices
- * to exec->vtx.copied.
- */
- vbo_exec_wrap_buffers( exec );
-
- if (unlikely(exec->vtx.copied.nr)) {
- /* We're in the middle of a primitive, keep the old vertex
- * format around to be able to translate the copied vertices to
- * the new format.
- */
- memcpy(old_attrptr, exec->vtx.attrptr, sizeof(old_attrptr));
- }
-
- if (unlikely(oldsz)) {
- /* Do a COPY_TO_CURRENT to ensure back-copying works for the
- * case when the attribute already exists in the vertex and is
- * having its size increased.
- */
- vbo_exec_copy_to_current( exec );
- }
-
- /* Heuristic: Attempt to isolate attributes received outside
- * begin/end so that they don't bloat the vertices.
- */
- if (ctx->Driver.CurrentExecPrimitive == PRIM_OUTSIDE_BEGIN_END &&
- !oldsz && lastcount > 8 && exec->vtx.vertex_size) {
- vbo_exec_copy_to_current( exec );
- reset_attrfv( exec );
- }
-
- /* Fix up sizes:
- */
- exec->vtx.attrsz[attr] = newsz;
- exec->vtx.vertex_size += newsz - oldsz;
- exec->vtx.max_vert = ((VBO_VERT_BUFFER_SIZE - exec->vtx.buffer_used) /
- (exec->vtx.vertex_size * sizeof(GLfloat)));
- exec->vtx.vert_count = 0;
- exec->vtx.buffer_ptr = exec->vtx.buffer_map;
-
- if (unlikely(oldsz)) {
- /* Size changed, recalculate all the attrptr[] values
- */
- GLfloat *tmp = exec->vtx.vertex;
-
- for (i = 0 ; i < VBO_ATTRIB_MAX ; i++) {
- if (exec->vtx.attrsz[i]) {
- exec->vtx.attrptr[i] = tmp;
- tmp += exec->vtx.attrsz[i];
- }
- else
- exec->vtx.attrptr[i] = NULL; /* will not be dereferenced */
- }
-
- /* Copy from current to repopulate the vertex with correct
- * values.
- */
- vbo_exec_copy_from_current( exec );
-
- } else {
- /* Just have to append the new attribute at the end */
- exec->vtx.attrptr[attr] = exec->vtx.vertex +
- exec->vtx.vertex_size - newsz;
- }
-
- /* Replay stored vertices to translate them
- * to new format here.
- *
- * -- No need to replay - just copy piecewise
- */
- if (unlikely(exec->vtx.copied.nr)) {
- GLfloat *data = exec->vtx.copied.buffer;
- GLfloat *dest = exec->vtx.buffer_ptr;
- GLuint j;
-
- assert(exec->vtx.buffer_ptr == exec->vtx.buffer_map);
-
- for (i = 0 ; i < exec->vtx.copied.nr ; i++) {
- for (j = 0 ; j < VBO_ATTRIB_MAX ; j++) {
- GLuint sz = exec->vtx.attrsz[j];
-
- if (sz) {
- GLint old_offset = old_attrptr[j] - exec->vtx.vertex;
- GLint new_offset = exec->vtx.attrptr[j] - exec->vtx.vertex;
-
- if (j == attr) {
- if (oldsz) {
- GLfloat tmp[4];
- COPY_CLEAN_4V(tmp, oldsz, data + old_offset);
- COPY_SZ_4V(dest + new_offset, newsz, tmp);
- } else {
- GLfloat *current = (GLfloat *)vbo->currval[j].Ptr;
- COPY_SZ_4V(dest + new_offset, sz, current);
- }
- }
- else {
- COPY_SZ_4V(dest + new_offset, sz, data + old_offset);
- }
- }
- }
-
- data += old_vtx_size;
- dest += exec->vtx.vertex_size;
- }
-
- exec->vtx.buffer_ptr = dest;
- exec->vtx.vert_count += exec->vtx.copied.nr;
- exec->vtx.copied.nr = 0;
- }
-}
-
-
-static void vbo_exec_fixup_vertex( struct gl_context *ctx,
- GLuint attr, GLuint sz )
-{
- struct vbo_exec_context *exec = &vbo_context(ctx)->exec;
- int i;
-
- if (sz > exec->vtx.attrsz[attr]) {
- /* New size is larger. Need to flush existing vertices and get
- * an enlarged vertex format.
- */
- vbo_exec_wrap_upgrade_vertex( exec, attr, sz );
- }
- else if (sz < exec->vtx.active_sz[attr]) {
- static const GLfloat id[4] = { 0, 0, 0, 1 };
-
- /* New size is smaller - just need to fill in some
- * zeros. Don't need to flush or wrap.
- */
- for (i = sz ; i <= exec->vtx.attrsz[attr] ; i++)
- exec->vtx.attrptr[attr][i-1] = id[i-1];
- }
-
- exec->vtx.active_sz[attr] = sz;
-
- /* Does setting NeedFlush belong here? Necessitates resetting
- * vtxfmt on each flush (otherwise flags won't get reset
- * afterwards).
- */
- if (attr == 0)
- exec->ctx->Driver.NeedFlush |= FLUSH_STORED_VERTICES;
-}
-
-
-/*
- */
-#define ATTR( A, N, V0, V1, V2, V3 ) \
-do { \
- struct vbo_exec_context *exec = &vbo_context(ctx)->exec; \
- \
- if (unlikely(!(exec->ctx->Driver.NeedFlush & FLUSH_UPDATE_CURRENT))) \
- ctx->Driver.BeginVertices( ctx ); \
- if (unlikely(exec->vtx.active_sz[A] != N)) \
- vbo_exec_fixup_vertex(ctx, A, N); \
- \
- { \
- GLfloat *dest = exec->vtx.attrptr[A]; \
- if (N>0) dest[0] = V0; \
- if (N>1) dest[1] = V1; \
- if (N>2) dest[2] = V2; \
- if (N>3) dest[3] = V3; \
- } \
- \
- if ((A) == 0) { \
- GLuint i; \
- \
- for (i = 0; i < exec->vtx.vertex_size; i++) \
- exec->vtx.buffer_ptr[i] = exec->vtx.vertex[i]; \
- \
- exec->vtx.buffer_ptr += exec->vtx.vertex_size; \
- exec->ctx->Driver.NeedFlush |= FLUSH_STORED_VERTICES; \
- \
- if (++exec->vtx.vert_count >= exec->vtx.max_vert) \
- vbo_exec_vtx_wrap( exec ); \
- } \
-} while (0)
-
-
-#define ERROR() _mesa_error( ctx, GL_INVALID_ENUM, __FUNCTION__ )
-#define TAG(x) vbo_##x
-
-#include "vbo_attrib_tmp.h"
-
-
-#if FEATURE_beginend
-
-
-#if FEATURE_evaluators
-
-static void GLAPIENTRY vbo_exec_EvalCoord1f( GLfloat u )
-{
- GET_CURRENT_CONTEXT( ctx );
- struct vbo_exec_context *exec = &vbo_context(ctx)->exec;
-
- {
- GLint i;
- if (exec->eval.recalculate_maps)
- vbo_exec_eval_update( exec );
-
- for (i = 0; i <= VBO_ATTRIB_TEX7; i++) {
- if (exec->eval.map1[i].map)
- if (exec->vtx.active_sz[i] != exec->eval.map1[i].sz)
- vbo_exec_fixup_vertex( ctx, i, exec->eval.map1[i].sz );
- }
- }
-
-
- memcpy( exec->vtx.copied.buffer, exec->vtx.vertex,
- exec->vtx.vertex_size * sizeof(GLfloat));
-
- vbo_exec_do_EvalCoord1f( exec, u );
-
- memcpy( exec->vtx.vertex, exec->vtx.copied.buffer,
- exec->vtx.vertex_size * sizeof(GLfloat));
-}
-
-static void GLAPIENTRY vbo_exec_EvalCoord2f( GLfloat u, GLfloat v )
-{
- GET_CURRENT_CONTEXT( ctx );
- struct vbo_exec_context *exec = &vbo_context(ctx)->exec;
-
- {
- GLint i;
- if (exec->eval.recalculate_maps)
- vbo_exec_eval_update( exec );
-
- for (i = 0; i <= VBO_ATTRIB_TEX7; i++) {
- if (exec->eval.map2[i].map)
- if (exec->vtx.active_sz[i] != exec->eval.map2[i].sz)
- vbo_exec_fixup_vertex( ctx, i, exec->eval.map2[i].sz );
- }
-
- if (ctx->Eval.AutoNormal)
- if (exec->vtx.active_sz[VBO_ATTRIB_NORMAL] != 3)
- vbo_exec_fixup_vertex( ctx, VBO_ATTRIB_NORMAL, 3 );
- }
-
- memcpy( exec->vtx.copied.buffer, exec->vtx.vertex,
- exec->vtx.vertex_size * sizeof(GLfloat));
-
- vbo_exec_do_EvalCoord2f( exec, u, v );
-
- memcpy( exec->vtx.vertex, exec->vtx.copied.buffer,
- exec->vtx.vertex_size * sizeof(GLfloat));
-}
-
-static void GLAPIENTRY vbo_exec_EvalCoord1fv( const GLfloat *u )
-{
- vbo_exec_EvalCoord1f( u[0] );
-}
-
-static void GLAPIENTRY vbo_exec_EvalCoord2fv( const GLfloat *u )
-{
- vbo_exec_EvalCoord2f( u[0], u[1] );
-}
-
-static void GLAPIENTRY vbo_exec_EvalPoint1( GLint i )
-{
- GET_CURRENT_CONTEXT( ctx );
- GLfloat du = ((ctx->Eval.MapGrid1u2 - ctx->Eval.MapGrid1u1) /
- (GLfloat) ctx->Eval.MapGrid1un);
- GLfloat u = i * du + ctx->Eval.MapGrid1u1;
-
- vbo_exec_EvalCoord1f( u );
-}
-
-
-static void GLAPIENTRY vbo_exec_EvalPoint2( GLint i, GLint j )
-{
- GET_CURRENT_CONTEXT( ctx );
- GLfloat du = ((ctx->Eval.MapGrid2u2 - ctx->Eval.MapGrid2u1) /
- (GLfloat) ctx->Eval.MapGrid2un);
- GLfloat dv = ((ctx->Eval.MapGrid2v2 - ctx->Eval.MapGrid2v1) /
- (GLfloat) ctx->Eval.MapGrid2vn);
- GLfloat u = i * du + ctx->Eval.MapGrid2u1;
- GLfloat v = j * dv + ctx->Eval.MapGrid2v1;
-
- vbo_exec_EvalCoord2f( u, v );
-}
-
-/* use noop eval mesh */
-#define vbo_exec_EvalMesh1 _mesa_noop_EvalMesh1
-#define vbo_exec_EvalMesh2 _mesa_noop_EvalMesh2
-
-#endif /* FEATURE_evaluators */
-
-
-/**
- * Called via glBegin.
- */
-static void GLAPIENTRY vbo_exec_Begin( GLenum mode )
-{
- GET_CURRENT_CONTEXT( ctx );
-
- if (ctx->Driver.CurrentExecPrimitive == PRIM_OUTSIDE_BEGIN_END) {
- struct vbo_exec_context *exec = &vbo_context(ctx)->exec;
- int i;
-
- if (ctx->NewState) {
- _mesa_update_state( ctx );
-
- CALL_Begin(ctx->Exec, (mode));
- return;
- }
-
- if (!_mesa_valid_to_render(ctx, "glBegin")) {
- return;
- }
-
- /* Heuristic: attempt to isolate attributes occuring outside
- * begin/end pairs.
- */
- if (exec->vtx.vertex_size && !exec->vtx.attrsz[0])
- vbo_exec_FlushVertices_internal( ctx, GL_FALSE );
-
- i = exec->vtx.prim_count++;
- exec->vtx.prim[i].mode = mode;
- exec->vtx.prim[i].begin = 1;
- exec->vtx.prim[i].end = 0;
- exec->vtx.prim[i].indexed = 0;
- exec->vtx.prim[i].weak = 0;
- exec->vtx.prim[i].pad = 0;
- exec->vtx.prim[i].start = exec->vtx.vert_count;
- exec->vtx.prim[i].count = 0;
- exec->vtx.prim[i].num_instances = 1;
-
- ctx->Driver.CurrentExecPrimitive = mode;
- }
- else
- _mesa_error( ctx, GL_INVALID_OPERATION, "glBegin" );
-
-}
-
-
-/**
- * Called via glEnd.
- */
-static void GLAPIENTRY vbo_exec_End( void )
-{
- GET_CURRENT_CONTEXT( ctx );
-
- if (ctx->Driver.CurrentExecPrimitive != PRIM_OUTSIDE_BEGIN_END) {
- struct vbo_exec_context *exec = &vbo_context(ctx)->exec;
- int idx = exec->vtx.vert_count;
- int i = exec->vtx.prim_count - 1;
-
- exec->vtx.prim[i].end = 1;
- exec->vtx.prim[i].count = idx - exec->vtx.prim[i].start;
-
- ctx->Driver.CurrentExecPrimitive = PRIM_OUTSIDE_BEGIN_END;
-
- if (exec->vtx.prim_count == VBO_MAX_PRIM)
- vbo_exec_vtx_flush( exec, GL_FALSE );
- }
- else
- _mesa_error( ctx, GL_INVALID_OPERATION, "glEnd" );
-}
-
-
-/**
- * Called via glPrimitiveRestartNV()
- */
-static void GLAPIENTRY
-vbo_exec_PrimitiveRestartNV(void)
-{
- GLenum curPrim;
- GET_CURRENT_CONTEXT( ctx );
-
- curPrim = ctx->Driver.CurrentExecPrimitive;
-
- if (curPrim == PRIM_OUTSIDE_BEGIN_END) {
- _mesa_error( ctx, GL_INVALID_OPERATION, "glPrimitiveRestartNV" );
- }
- else {
- vbo_exec_End();
- vbo_exec_Begin(curPrim);
- }
-}
-
-
-
-static void vbo_exec_vtxfmt_init( struct vbo_exec_context *exec )
-{
- GLvertexformat *vfmt = &exec->vtxfmt;
-
- _MESA_INIT_ARRAYELT_VTXFMT(vfmt, _ae_);
-
- vfmt->Begin = vbo_exec_Begin;
- vfmt->End = vbo_exec_End;
- vfmt->PrimitiveRestartNV = vbo_exec_PrimitiveRestartNV;
-
- _MESA_INIT_DLIST_VTXFMT(vfmt, _mesa_);
- _MESA_INIT_EVAL_VTXFMT(vfmt, vbo_exec_);
-
- vfmt->Rectf = _mesa_noop_Rectf;
-
- /* from attrib_tmp.h:
- */
- vfmt->Color3f = vbo_Color3f;
- vfmt->Color3fv = vbo_Color3fv;
- vfmt->Color4f = vbo_Color4f;
- vfmt->Color4fv = vbo_Color4fv;
- vfmt->FogCoordfEXT = vbo_FogCoordfEXT;
- vfmt->FogCoordfvEXT = vbo_FogCoordfvEXT;
- vfmt->MultiTexCoord1fARB = vbo_MultiTexCoord1f;
- vfmt->MultiTexCoord1fvARB = vbo_MultiTexCoord1fv;
- vfmt->MultiTexCoord2fARB = vbo_MultiTexCoord2f;
- vfmt->MultiTexCoord2fvARB = vbo_MultiTexCoord2fv;
- vfmt->MultiTexCoord3fARB = vbo_MultiTexCoord3f;
- vfmt->MultiTexCoord3fvARB = vbo_MultiTexCoord3fv;
- vfmt->MultiTexCoord4fARB = vbo_MultiTexCoord4f;
- vfmt->MultiTexCoord4fvARB = vbo_MultiTexCoord4fv;
- vfmt->Normal3f = vbo_Normal3f;
- vfmt->Normal3fv = vbo_Normal3fv;
- vfmt->SecondaryColor3fEXT = vbo_SecondaryColor3fEXT;
- vfmt->SecondaryColor3fvEXT = vbo_SecondaryColor3fvEXT;
- vfmt->TexCoord1f = vbo_TexCoord1f;
- vfmt->TexCoord1fv = vbo_TexCoord1fv;
- vfmt->TexCoord2f = vbo_TexCoord2f;
- vfmt->TexCoord2fv = vbo_TexCoord2fv;
- vfmt->TexCoord3f = vbo_TexCoord3f;
- vfmt->TexCoord3fv = vbo_TexCoord3fv;
- vfmt->TexCoord4f = vbo_TexCoord4f;
- vfmt->TexCoord4fv = vbo_TexCoord4fv;
- vfmt->Vertex2f = vbo_Vertex2f;
- vfmt->Vertex2fv = vbo_Vertex2fv;
- vfmt->Vertex3f = vbo_Vertex3f;
- vfmt->Vertex3fv = vbo_Vertex3fv;
- vfmt->Vertex4f = vbo_Vertex4f;
- vfmt->Vertex4fv = vbo_Vertex4fv;
-
- vfmt->VertexAttrib1fARB = vbo_VertexAttrib1fARB;
- vfmt->VertexAttrib1fvARB = vbo_VertexAttrib1fvARB;
- vfmt->VertexAttrib2fARB = vbo_VertexAttrib2fARB;
- vfmt->VertexAttrib2fvARB = vbo_VertexAttrib2fvARB;
- vfmt->VertexAttrib3fARB = vbo_VertexAttrib3fARB;
- vfmt->VertexAttrib3fvARB = vbo_VertexAttrib3fvARB;
- vfmt->VertexAttrib4fARB = vbo_VertexAttrib4fARB;
- vfmt->VertexAttrib4fvARB = vbo_VertexAttrib4fvARB;
-
- vfmt->VertexAttrib1fNV = vbo_VertexAttrib1fNV;
- vfmt->VertexAttrib1fvNV = vbo_VertexAttrib1fvNV;
- vfmt->VertexAttrib2fNV = vbo_VertexAttrib2fNV;
- vfmt->VertexAttrib2fvNV = vbo_VertexAttrib2fvNV;
- vfmt->VertexAttrib3fNV = vbo_VertexAttrib3fNV;
- vfmt->VertexAttrib3fvNV = vbo_VertexAttrib3fvNV;
- vfmt->VertexAttrib4fNV = vbo_VertexAttrib4fNV;
- vfmt->VertexAttrib4fvNV = vbo_VertexAttrib4fvNV;
-
- /* integer-valued */
- vfmt->VertexAttribI1i = vbo_VertexAttribI1i;
- vfmt->VertexAttribI2i = vbo_VertexAttribI2i;
- vfmt->VertexAttribI3i = vbo_VertexAttribI3i;
- vfmt->VertexAttribI4i = vbo_VertexAttribI4i;
- vfmt->VertexAttribI2iv = vbo_VertexAttribI2iv;
- vfmt->VertexAttribI3iv = vbo_VertexAttribI3iv;
- vfmt->VertexAttribI4iv = vbo_VertexAttribI4iv;
-
- /* unsigned integer-valued */
- vfmt->VertexAttribI1ui = vbo_VertexAttribI1ui;
- vfmt->VertexAttribI2ui = vbo_VertexAttribI2ui;
- vfmt->VertexAttribI3ui = vbo_VertexAttribI3ui;
- vfmt->VertexAttribI4ui = vbo_VertexAttribI4ui;
- vfmt->VertexAttribI2uiv = vbo_VertexAttribI2uiv;
- vfmt->VertexAttribI3uiv = vbo_VertexAttribI3uiv;
- vfmt->VertexAttribI4uiv = vbo_VertexAttribI4uiv;
-
- vfmt->Materialfv = vbo_Materialfv;
-
- vfmt->EdgeFlag = vbo_EdgeFlag;
- vfmt->Indexf = vbo_Indexf;
- vfmt->Indexfv = vbo_Indexfv;
-
-}
-
-
-#else /* FEATURE_beginend */
-
-
-static void vbo_exec_vtxfmt_init( struct vbo_exec_context *exec )
-{
- /* silence warnings */
- (void) vbo_Color3f;
- (void) vbo_Color3fv;
- (void) vbo_Color4f;
- (void) vbo_Color4fv;
- (void) vbo_FogCoordfEXT;
- (void) vbo_FogCoordfvEXT;
- (void) vbo_MultiTexCoord1f;
- (void) vbo_MultiTexCoord1fv;
- (void) vbo_MultiTexCoord2f;
- (void) vbo_MultiTexCoord2fv;
- (void) vbo_MultiTexCoord3f;
- (void) vbo_MultiTexCoord3fv;
- (void) vbo_MultiTexCoord4f;
- (void) vbo_MultiTexCoord4fv;
- (void) vbo_Normal3f;
- (void) vbo_Normal3fv;
- (void) vbo_SecondaryColor3fEXT;
- (void) vbo_SecondaryColor3fvEXT;
- (void) vbo_TexCoord1f;
- (void) vbo_TexCoord1fv;
- (void) vbo_TexCoord2f;
- (void) vbo_TexCoord2fv;
- (void) vbo_TexCoord3f;
- (void) vbo_TexCoord3fv;
- (void) vbo_TexCoord4f;
- (void) vbo_TexCoord4fv;
- (void) vbo_Vertex2f;
- (void) vbo_Vertex2fv;
- (void) vbo_Vertex3f;
- (void) vbo_Vertex3fv;
- (void) vbo_Vertex4f;
- (void) vbo_Vertex4fv;
-
- (void) vbo_VertexAttrib1fARB;
- (void) vbo_VertexAttrib1fvARB;
- (void) vbo_VertexAttrib2fARB;
- (void) vbo_VertexAttrib2fvARB;
- (void) vbo_VertexAttrib3fARB;
- (void) vbo_VertexAttrib3fvARB;
- (void) vbo_VertexAttrib4fARB;
- (void) vbo_VertexAttrib4fvARB;
-
- (void) vbo_VertexAttrib1fNV;
- (void) vbo_VertexAttrib1fvNV;
- (void) vbo_VertexAttrib2fNV;
- (void) vbo_VertexAttrib2fvNV;
- (void) vbo_VertexAttrib3fNV;
- (void) vbo_VertexAttrib3fvNV;
- (void) vbo_VertexAttrib4fNV;
- (void) vbo_VertexAttrib4fvNV;
-
- (void) vbo_Materialfv;
-
- (void) vbo_EdgeFlag;
- (void) vbo_Indexf;
- (void) vbo_Indexfv;
-}
-
-
-#endif /* FEATURE_beginend */
-
-
-/**
- * Tell the VBO module to use a real OpenGL vertex buffer object to
- * store accumulated immediate-mode vertex data.
- * This replaces the malloced buffer which was created in
- * vb_exec_vtx_init() below.
- */
-void vbo_use_buffer_objects(struct gl_context *ctx)
-{
- struct vbo_exec_context *exec = &vbo_context(ctx)->exec;
- /* Any buffer name but 0 can be used here since this bufferobj won't
- * go into the bufferobj hashtable.
- */
- GLuint bufName = IMM_BUFFER_NAME;
- GLenum target = GL_ARRAY_BUFFER_ARB;
- GLenum usage = GL_STREAM_DRAW_ARB;
- GLsizei size = VBO_VERT_BUFFER_SIZE;
-
- /* Make sure this func is only used once */
- assert(exec->vtx.bufferobj == ctx->Shared->NullBufferObj);
- if (exec->vtx.buffer_map) {
- _mesa_align_free(exec->vtx.buffer_map);
- exec->vtx.buffer_map = NULL;
- exec->vtx.buffer_ptr = NULL;
- }
-
- /* Allocate a real buffer object now */
- _mesa_reference_buffer_object(ctx, &exec->vtx.bufferobj, NULL);
- exec->vtx.bufferobj = ctx->Driver.NewBufferObject(ctx, bufName, target);
- ctx->Driver.BufferData(ctx, target, size, NULL, usage, exec->vtx.bufferobj);
-}
-
-
-
-void vbo_exec_vtx_init( struct vbo_exec_context *exec )
-{
- struct gl_context *ctx = exec->ctx;
- struct vbo_context *vbo = vbo_context(ctx);
- GLuint i;
-
- /* Allocate a buffer object. Will just reuse this object
- * continuously, unless vbo_use_buffer_objects() is called to enable
- * use of real VBOs.
- */
- _mesa_reference_buffer_object(ctx,
- &exec->vtx.bufferobj,
- ctx->Shared->NullBufferObj);
-
- ASSERT(!exec->vtx.buffer_map);
- exec->vtx.buffer_map = (GLfloat *)_mesa_align_malloc(VBO_VERT_BUFFER_SIZE, 64);
- exec->vtx.buffer_ptr = exec->vtx.buffer_map;
-
- vbo_exec_vtxfmt_init( exec );
-
- /* Hook our functions into the dispatch table.
- */
- _mesa_install_exec_vtxfmt( exec->ctx, &exec->vtxfmt );
-
- for (i = 0 ; i < VBO_ATTRIB_MAX ; i++) {
- ASSERT(i < Elements(exec->vtx.attrsz));
- exec->vtx.attrsz[i] = 0;
- ASSERT(i < Elements(exec->vtx.active_sz));
- exec->vtx.active_sz[i] = 0;
- }
- for (i = 0 ; i < VERT_ATTRIB_MAX; i++) {
- ASSERT(i < Elements(exec->vtx.inputs));
- ASSERT(i < Elements(exec->vtx.arrays));
- exec->vtx.inputs[i] = &exec->vtx.arrays[i];
- }
-
- {
- struct gl_client_array *arrays = exec->vtx.arrays;
- unsigned i;
-
- memcpy(arrays, vbo->legacy_currval, 16 * sizeof(arrays[0]));
- memcpy(arrays + 16, vbo->generic_currval, 16 * sizeof(arrays[0]));
-
- for (i = 0; i < 16; ++i) {
- arrays[i ].BufferObj = NULL;
- arrays[i + 16].BufferObj = NULL;
- _mesa_reference_buffer_object(ctx, &arrays[i ].BufferObj,
- vbo->legacy_currval[i].BufferObj);
- _mesa_reference_buffer_object(ctx, &arrays[i + 16].BufferObj,
- vbo->generic_currval[i].BufferObj);
- }
- }
-
- exec->vtx.vertex_size = 0;
-}
-
-
-void vbo_exec_vtx_destroy( struct vbo_exec_context *exec )
-{
- /* using a real VBO for vertex data */
- struct gl_context *ctx = exec->ctx;
- unsigned i;
-
- /* True VBOs should already be unmapped
- */
- if (exec->vtx.buffer_map) {
- ASSERT(exec->vtx.bufferobj->Name == 0 ||
- exec->vtx.bufferobj->Name == IMM_BUFFER_NAME);
- if (exec->vtx.bufferobj->Name == 0) {
- _mesa_align_free(exec->vtx.buffer_map);
- exec->vtx.buffer_map = NULL;
- exec->vtx.buffer_ptr = NULL;
- }
- }
-
- /* Drop any outstanding reference to the vertex buffer
- */
- for (i = 0; i < Elements(exec->vtx.arrays); i++) {
- _mesa_reference_buffer_object(ctx,
- &exec->vtx.arrays[i].BufferObj,
- NULL);
- }
-
- /* Free the vertex buffer. Unmap first if needed.
- */
- if (_mesa_bufferobj_mapped(exec->vtx.bufferobj)) {
- ctx->Driver.UnmapBuffer(ctx, GL_ARRAY_BUFFER, exec->vtx.bufferobj);
- }
- _mesa_reference_buffer_object(ctx, &exec->vtx.bufferobj, NULL);
-}
-
-void vbo_exec_BeginVertices( struct gl_context *ctx )
-{
- struct vbo_exec_context *exec = &vbo_context(ctx)->exec;
- if (0) printf("%s\n", __FUNCTION__);
- vbo_exec_vtx_map( exec );
-
- assert((exec->ctx->Driver.NeedFlush & FLUSH_UPDATE_CURRENT) == 0);
- exec->ctx->Driver.NeedFlush |= FLUSH_UPDATE_CURRENT;
-}
-
-void vbo_exec_FlushVertices_internal( struct gl_context *ctx, GLboolean unmap )
-{
- struct vbo_exec_context *exec = &vbo_context(ctx)->exec;
-
- if (exec->vtx.vert_count || unmap) {
- vbo_exec_vtx_flush( exec, unmap );
- }
-
- if (exec->vtx.vertex_size) {
- vbo_exec_copy_to_current( exec );
- reset_attrfv( exec );
- }
-}
-
-
-/**
- * \param flags bitmask of FLUSH_STORED_VERTICES, FLUSH_UPDATE_CURRENT
- */
-void vbo_exec_FlushVertices( struct gl_context *ctx, GLuint flags )
-{
- struct vbo_exec_context *exec = &vbo_context(ctx)->exec;
-
-#ifdef DEBUG
- /* debug check: make sure we don't get called recursively */
- exec->flush_call_depth++;
- assert(exec->flush_call_depth == 1);
-#endif
-
- if (0) printf("%s\n", __FUNCTION__);
-
- if (exec->ctx->Driver.CurrentExecPrimitive != PRIM_OUTSIDE_BEGIN_END) {
- if (0) printf("%s - inside begin/end\n", __FUNCTION__);
-#ifdef DEBUG
- exec->flush_call_depth--;
- assert(exec->flush_call_depth == 0);
-#endif
- return;
- }
-
- vbo_exec_FlushVertices_internal( ctx, GL_TRUE );
-
- /* Need to do this to ensure BeginVertices gets called again:
- */
- if (exec->ctx->Driver.NeedFlush & FLUSH_UPDATE_CURRENT)
- exec->ctx->Driver.NeedFlush &= ~FLUSH_UPDATE_CURRENT;
-
- exec->ctx->Driver.NeedFlush &= ~flags;
-
-#ifdef DEBUG
- exec->flush_call_depth--;
- assert(exec->flush_call_depth == 0);
-#endif
-}
-
-
-static void reset_attrfv( struct vbo_exec_context *exec )
-{
- GLuint i;
-
- for (i = 0 ; i < VBO_ATTRIB_MAX ; i++) {
- exec->vtx.attrsz[i] = 0;
- exec->vtx.active_sz[i] = 0;
- }
-
- exec->vtx.vertex_size = 0;
-}
-
-
-void GLAPIENTRY
-_es_Color4f(GLfloat r, GLfloat g, GLfloat b, GLfloat a)
-{
- vbo_Color4f(r, g, b, a);
-}
-
-
-void GLAPIENTRY
-_es_Normal3f(GLfloat x, GLfloat y, GLfloat z)
-{
- vbo_Normal3f(x, y, z);
-}
-
-
-void GLAPIENTRY
-_es_MultiTexCoord4f(GLenum target, GLfloat s, GLfloat t, GLfloat r, GLfloat q)
-{
- vbo_MultiTexCoord4f(target, s, t, r, q);
-}
-
-
-void GLAPIENTRY
-_es_Materialfv(GLenum face, GLenum pname, const GLfloat *params)
-{
- vbo_Materialfv(face, pname, params);
-}
-
-
-void GLAPIENTRY
-_es_Materialf(GLenum face, GLenum pname, GLfloat param)
-{
- GLfloat p[4];
- p[0] = param;
- p[1] = p[2] = p[3] = 0.0F;
- vbo_Materialfv(face, pname, p);
-}
-
-
-/**
- * A special version of glVertexAttrib4f that does not treat index 0 as
- * VBO_ATTRIB_POS.
- */
-static void
-VertexAttrib4f_nopos(GLuint index, GLfloat x, GLfloat y, GLfloat z, GLfloat w)
-{
- GET_CURRENT_CONTEXT(ctx);
- if (index < MAX_VERTEX_GENERIC_ATTRIBS)
- ATTR(VBO_ATTRIB_GENERIC0 + index, 4, x, y, z, w);
- else
- ERROR();
-}
-
-void GLAPIENTRY
-_es_VertexAttrib4f(GLuint index, GLfloat x, GLfloat y, GLfloat z, GLfloat w)
-{
- VertexAttrib4f_nopos(index, x, y, z, w);
-}
-
-
-void GLAPIENTRY
-_es_VertexAttrib1f(GLuint indx, GLfloat x)
-{
- VertexAttrib4f_nopos(indx, x, 0.0f, 0.0f, 1.0f);
-}
-
-
-void GLAPIENTRY
-_es_VertexAttrib1fv(GLuint indx, const GLfloat* values)
-{
- VertexAttrib4f_nopos(indx, values[0], 0.0f, 0.0f, 1.0f);
-}
-
-
-void GLAPIENTRY
-_es_VertexAttrib2f(GLuint indx, GLfloat x, GLfloat y)
-{
- VertexAttrib4f_nopos(indx, x, y, 0.0f, 1.0f);
-}
-
-
-void GLAPIENTRY
-_es_VertexAttrib2fv(GLuint indx, const GLfloat* values)
-{
- VertexAttrib4f_nopos(indx, values[0], values[1], 0.0f, 1.0f);
-}
-
-
-void GLAPIENTRY
-_es_VertexAttrib3f(GLuint indx, GLfloat x, GLfloat y, GLfloat z)
-{
- VertexAttrib4f_nopos(indx, x, y, z, 1.0f);
-}
-
-
-void GLAPIENTRY
-_es_VertexAttrib3fv(GLuint indx, const GLfloat* values)
-{
- VertexAttrib4f_nopos(indx, values[0], values[1], values[2], 1.0f);
-}
-
-
-void GLAPIENTRY
-_es_VertexAttrib4fv(GLuint indx, const GLfloat* values)
-{
- VertexAttrib4f_nopos(indx, values[0], values[1], values[2], values[3]);
-}
+/************************************************************************** + +Copyright 2002-2008 Tungsten Graphics Inc., Cedar Park, Texas. + +All Rights Reserved. + +Permission is hereby granted, free of charge, to any person obtaining a +copy of this software and associated documentation files (the "Software"), +to deal in the Software without restriction, including without limitation +on the rights to use, copy, modify, merge, publish, distribute, sub +license, and/or sell copies of the Software, and to permit persons to whom +the Software is furnished to do so, subject to the following conditions: + +The above copyright notice and this permission notice (including the next +paragraph) shall be included in all copies or substantial portions of the +Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL +TUNGSTEN GRAPHICS AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM, +DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR +OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE +USE OR OTHER DEALINGS IN THE SOFTWARE. + +**************************************************************************/ + +/* + * Authors: + * Keith Whitwell <keith@tungstengraphics.com> + */ + +#include "main/glheader.h" +#include "main/bufferobj.h" +#include "main/context.h" +#include "main/macros.h" +#include "main/mfeatures.h" +#include "main/vtxfmt.h" +#include "main/dlist.h" +#include "main/eval.h" +#include "main/state.h" +#include "main/light.h" +#include "main/api_arrayelt.h" +#include "main/api_noop.h" +#include "main/dispatch.h" + +#include "vbo_context.h" + +#ifdef ERROR +#undef ERROR +#endif + + +/** ID/name for immediate-mode VBO */ +#define IMM_BUFFER_NAME 0xaabbccdd + + +static void reset_attrfv( struct vbo_exec_context *exec ); + + +/** + * Close off the last primitive, execute the buffer, restart the + * primitive. + */ +static void vbo_exec_wrap_buffers( struct vbo_exec_context *exec ) +{ + if (exec->vtx.prim_count == 0) { + exec->vtx.copied.nr = 0; + exec->vtx.vert_count = 0; + exec->vtx.buffer_ptr = exec->vtx.buffer_map; + } + else { + GLuint last_begin = exec->vtx.prim[exec->vtx.prim_count-1].begin; + GLuint last_count; + + if (exec->ctx->Driver.CurrentExecPrimitive != PRIM_OUTSIDE_BEGIN_END) { + GLint i = exec->vtx.prim_count - 1; + assert(i >= 0); + exec->vtx.prim[i].count = (exec->vtx.vert_count - + exec->vtx.prim[i].start); + } + + last_count = exec->vtx.prim[exec->vtx.prim_count-1].count; + + /* Execute the buffer and save copied vertices. + */ + if (exec->vtx.vert_count) + vbo_exec_vtx_flush( exec, GL_FALSE ); + else { + exec->vtx.prim_count = 0; + exec->vtx.copied.nr = 0; + } + + /* Emit a glBegin to start the new list. + */ + assert(exec->vtx.prim_count == 0); + + if (exec->ctx->Driver.CurrentExecPrimitive != PRIM_OUTSIDE_BEGIN_END) { + exec->vtx.prim[0].mode = exec->ctx->Driver.CurrentExecPrimitive; + exec->vtx.prim[0].start = 0; + exec->vtx.prim[0].count = 0; + exec->vtx.prim_count++; + + if (exec->vtx.copied.nr == last_count) + exec->vtx.prim[0].begin = last_begin; + } + } +} + + +/** + * Deal with buffer wrapping where provoked by the vertex buffer + * filling up, as opposed to upgrade_vertex(). + */ +void vbo_exec_vtx_wrap( struct vbo_exec_context *exec ) +{ + GLfloat *data = exec->vtx.copied.buffer; + GLuint i; + + /* Run pipeline on current vertices, copy wrapped vertices + * to exec->vtx.copied. + */ + vbo_exec_wrap_buffers( exec ); + + /* Copy stored stored vertices to start of new list. + */ + assert(exec->vtx.max_vert - exec->vtx.vert_count > exec->vtx.copied.nr); + + for (i = 0 ; i < exec->vtx.copied.nr ; i++) { + memcpy( exec->vtx.buffer_ptr, data, + exec->vtx.vertex_size * sizeof(GLfloat)); + exec->vtx.buffer_ptr += exec->vtx.vertex_size; + data += exec->vtx.vertex_size; + exec->vtx.vert_count++; + } + + exec->vtx.copied.nr = 0; +} + + +/** + * Copy the active vertex's values to the ctx->Current fields. + */ +static void vbo_exec_copy_to_current( struct vbo_exec_context *exec ) +{ + struct gl_context *ctx = exec->ctx; + struct vbo_context *vbo = vbo_context(ctx); + GLuint i; + + for (i = VBO_ATTRIB_POS+1 ; i < VBO_ATTRIB_MAX ; i++) { + if (exec->vtx.attrsz[i]) { + /* Note: the exec->vtx.current[i] pointers point into the + * ctx->Current.Attrib and ctx->Light.Material.Attrib arrays. + */ + GLfloat *current = (GLfloat *)vbo->currval[i].Ptr; + GLfloat tmp[4]; + + COPY_CLEAN_4V(tmp, + exec->vtx.attrsz[i], + exec->vtx.attrptr[i]); + + if (memcmp(current, tmp, sizeof(tmp)) != 0) { + memcpy(current, tmp, sizeof(tmp)); + + /* Given that we explicitly state size here, there is no need + * for the COPY_CLEAN above, could just copy 16 bytes and be + * done. The only problem is when Mesa accesses ctx->Current + * directly. + */ + vbo->currval[i].Size = exec->vtx.attrsz[i]; + + /* This triggers rather too much recalculation of Mesa state + * that doesn't get used (eg light positions). + */ + if (i >= VBO_ATTRIB_MAT_FRONT_AMBIENT && + i <= VBO_ATTRIB_MAT_BACK_INDEXES) + ctx->NewState |= _NEW_LIGHT; + + ctx->NewState |= _NEW_CURRENT_ATTRIB; + } + } + } + + /* Colormaterial -- this kindof sucks. + */ + if (ctx->Light.ColorMaterialEnabled && + exec->vtx.attrsz[VBO_ATTRIB_COLOR0]) { + _mesa_update_color_material(ctx, + ctx->Current.Attrib[VBO_ATTRIB_COLOR0]); + } +} + + +/** + * Copy current vertex attribute values into the current vertex. + */ +static void +vbo_exec_copy_from_current(struct vbo_exec_context *exec) +{ + struct gl_context *ctx = exec->ctx; + struct vbo_context *vbo = vbo_context(ctx); + GLint i; + + for (i = VBO_ATTRIB_POS + 1; i < VBO_ATTRIB_MAX; i++) { + const GLfloat *current = (GLfloat *) vbo->currval[i].Ptr; + switch (exec->vtx.attrsz[i]) { + case 4: exec->vtx.attrptr[i][3] = current[3]; + case 3: exec->vtx.attrptr[i][2] = current[2]; + case 2: exec->vtx.attrptr[i][1] = current[1]; + case 1: exec->vtx.attrptr[i][0] = current[0]; + break; + } + } +} + + +/** + * Flush existing data, set new attrib size, replay copied vertices. + * This is called when we transition from a small vertex attribute size + * to a larger one. Ex: glTexCoord2f -> glTexCoord4f. + * We need to go back over the previous 2-component texcoords and insert + * zero and one values. + */ +static void +vbo_exec_wrap_upgrade_vertex(struct vbo_exec_context *exec, + GLuint attr, GLuint newSize ) +{ + struct gl_context *ctx = exec->ctx; + struct vbo_context *vbo = vbo_context(ctx); + const GLint lastcount = exec->vtx.vert_count; + GLfloat *old_attrptr[VBO_ATTRIB_MAX]; + const GLuint old_vtx_size = exec->vtx.vertex_size; /* floats per vertex */ + const GLuint oldSize = exec->vtx.attrsz[attr]; + GLuint i; + + /* Run pipeline on current vertices, copy wrapped vertices + * to exec->vtx.copied. + */ + vbo_exec_wrap_buffers( exec ); + + if (unlikely(exec->vtx.copied.nr)) { + /* We're in the middle of a primitive, keep the old vertex + * format around to be able to translate the copied vertices to + * the new format. + */ + memcpy(old_attrptr, exec->vtx.attrptr, sizeof(old_attrptr)); + } + + if (unlikely(oldSize)) { + /* Do a COPY_TO_CURRENT to ensure back-copying works for the + * case when the attribute already exists in the vertex and is + * having its size increased. + */ + vbo_exec_copy_to_current( exec ); + } + + /* Heuristic: Attempt to isolate attributes received outside + * begin/end so that they don't bloat the vertices. + */ + if (ctx->Driver.CurrentExecPrimitive == PRIM_OUTSIDE_BEGIN_END && + !oldSize && lastcount > 8 && exec->vtx.vertex_size) { + vbo_exec_copy_to_current( exec ); + reset_attrfv( exec ); + } + + /* Fix up sizes: + */ + exec->vtx.attrsz[attr] = newSize; + exec->vtx.vertex_size += newSize - oldSize; + exec->vtx.max_vert = ((VBO_VERT_BUFFER_SIZE - exec->vtx.buffer_used) / + (exec->vtx.vertex_size * sizeof(GLfloat))); + exec->vtx.vert_count = 0; + exec->vtx.buffer_ptr = exec->vtx.buffer_map; + + if (unlikely(oldSize)) { + /* Size changed, recalculate all the attrptr[] values + */ + GLfloat *tmp = exec->vtx.vertex; + + for (i = 0 ; i < VBO_ATTRIB_MAX ; i++) { + if (exec->vtx.attrsz[i]) { + exec->vtx.attrptr[i] = tmp; + tmp += exec->vtx.attrsz[i]; + } + else + exec->vtx.attrptr[i] = NULL; /* will not be dereferenced */ + } + + /* Copy from current to repopulate the vertex with correct + * values. + */ + vbo_exec_copy_from_current( exec ); + } + else { + /* Just have to append the new attribute at the end */ + exec->vtx.attrptr[attr] = exec->vtx.vertex + + exec->vtx.vertex_size - newSize; + } + + /* Replay stored vertices to translate them + * to new format here. + * + * -- No need to replay - just copy piecewise + */ + if (unlikely(exec->vtx.copied.nr)) { + GLfloat *data = exec->vtx.copied.buffer; + GLfloat *dest = exec->vtx.buffer_ptr; + GLuint j; + + assert(exec->vtx.buffer_ptr == exec->vtx.buffer_map); + + for (i = 0 ; i < exec->vtx.copied.nr ; i++) { + for (j = 0 ; j < VBO_ATTRIB_MAX ; j++) { + GLuint sz = exec->vtx.attrsz[j]; + + if (sz) { + GLint old_offset = old_attrptr[j] - exec->vtx.vertex; + GLint new_offset = exec->vtx.attrptr[j] - exec->vtx.vertex; + + if (j == attr) { + if (oldSize) { + GLfloat tmp[4]; + COPY_CLEAN_4V(tmp, oldSize, data + old_offset); + COPY_SZ_4V(dest + new_offset, newSize, tmp); + } else { + GLfloat *current = (GLfloat *)vbo->currval[j].Ptr; + COPY_SZ_4V(dest + new_offset, sz, current); + } + } + else { + COPY_SZ_4V(dest + new_offset, sz, data + old_offset); + } + } + } + + data += old_vtx_size; + dest += exec->vtx.vertex_size; + } + + exec->vtx.buffer_ptr = dest; + exec->vtx.vert_count += exec->vtx.copied.nr; + exec->vtx.copied.nr = 0; + } +} + + +/** + * This is when a vertex attribute transitions to a different size. + * For example, we saw a bunch of glTexCoord2f() calls and now we got a + * glTexCoord4f() call. We promote the array from size=2 to size=4. + */ +static void +vbo_exec_fixup_vertex(struct gl_context *ctx, GLuint attr, GLuint newSize) +{ + struct vbo_exec_context *exec = &vbo_context(ctx)->exec; + + if (newSize > exec->vtx.attrsz[attr]) { + /* New size is larger. Need to flush existing vertices and get + * an enlarged vertex format. + */ + vbo_exec_wrap_upgrade_vertex( exec, attr, newSize ); + } + else if (newSize < exec->vtx.active_sz[attr]) { + static const GLfloat id[4] = { 0, 0, 0, 1 }; + GLuint i; + + /* New size is smaller - just need to fill in some + * zeros. Don't need to flush or wrap. + */ + for (i = newSize; i <= exec->vtx.attrsz[attr]; i++) + exec->vtx.attrptr[attr][i-1] = id[i-1]; + } + + exec->vtx.active_sz[attr] = newSize; + + /* Does setting NeedFlush belong here? Necessitates resetting + * vtxfmt on each flush (otherwise flags won't get reset + * afterwards). + */ + if (attr == 0) + exec->ctx->Driver.NeedFlush |= FLUSH_STORED_VERTICES; +} + + +/** + * This macro is used to implement all the glVertex, glColor, glTexCoord, + * glVertexAttrib, etc functions. + */ +#define ATTR( A, N, V0, V1, V2, V3 ) \ +do { \ + struct vbo_exec_context *exec = &vbo_context(ctx)->exec; \ + \ + if (unlikely(!(ctx->Driver.NeedFlush & FLUSH_UPDATE_CURRENT))) \ + ctx->Driver.BeginVertices( ctx ); \ + \ + if (unlikely(exec->vtx.active_sz[A] != N)) \ + vbo_exec_fixup_vertex(ctx, A, N); \ + \ + { \ + GLfloat *dest = exec->vtx.attrptr[A]; \ + if (N>0) dest[0] = V0; \ + if (N>1) dest[1] = V1; \ + if (N>2) dest[2] = V2; \ + if (N>3) dest[3] = V3; \ + } \ + \ + if ((A) == 0) { \ + /* This is a glVertex call */ \ + GLuint i; \ + \ + for (i = 0; i < exec->vtx.vertex_size; i++) \ + exec->vtx.buffer_ptr[i] = exec->vtx.vertex[i]; \ + \ + exec->vtx.buffer_ptr += exec->vtx.vertex_size; \ + \ + /* Set FLUSH_STORED_VERTICES to indicate that there's now */ \ + /* something to draw (not just updating a color or texcoord).*/ \ + ctx->Driver.NeedFlush |= FLUSH_STORED_VERTICES; \ + \ + if (++exec->vtx.vert_count >= exec->vtx.max_vert) \ + vbo_exec_vtx_wrap( exec ); \ + } \ +} while (0) + + +#define ERROR() _mesa_error( ctx, GL_INVALID_ENUM, __FUNCTION__ ) +#define TAG(x) vbo_##x + +#include "vbo_attrib_tmp.h" + + +#if FEATURE_beginend + + +#if FEATURE_evaluators + +static void GLAPIENTRY vbo_exec_EvalCoord1f( GLfloat u ) +{ + GET_CURRENT_CONTEXT( ctx ); + struct vbo_exec_context *exec = &vbo_context(ctx)->exec; + + { + GLint i; + if (exec->eval.recalculate_maps) + vbo_exec_eval_update( exec ); + + for (i = 0; i <= VBO_ATTRIB_TEX7; i++) { + if (exec->eval.map1[i].map) + if (exec->vtx.active_sz[i] != exec->eval.map1[i].sz) + vbo_exec_fixup_vertex( ctx, i, exec->eval.map1[i].sz ); + } + } + + + memcpy( exec->vtx.copied.buffer, exec->vtx.vertex, + exec->vtx.vertex_size * sizeof(GLfloat)); + + vbo_exec_do_EvalCoord1f( exec, u ); + + memcpy( exec->vtx.vertex, exec->vtx.copied.buffer, + exec->vtx.vertex_size * sizeof(GLfloat)); +} + +static void GLAPIENTRY vbo_exec_EvalCoord2f( GLfloat u, GLfloat v ) +{ + GET_CURRENT_CONTEXT( ctx ); + struct vbo_exec_context *exec = &vbo_context(ctx)->exec; + + { + GLint i; + if (exec->eval.recalculate_maps) + vbo_exec_eval_update( exec ); + + for (i = 0; i <= VBO_ATTRIB_TEX7; i++) { + if (exec->eval.map2[i].map) + if (exec->vtx.active_sz[i] != exec->eval.map2[i].sz) + vbo_exec_fixup_vertex( ctx, i, exec->eval.map2[i].sz ); + } + + if (ctx->Eval.AutoNormal) + if (exec->vtx.active_sz[VBO_ATTRIB_NORMAL] != 3) + vbo_exec_fixup_vertex( ctx, VBO_ATTRIB_NORMAL, 3 ); + } + + memcpy( exec->vtx.copied.buffer, exec->vtx.vertex, + exec->vtx.vertex_size * sizeof(GLfloat)); + + vbo_exec_do_EvalCoord2f( exec, u, v ); + + memcpy( exec->vtx.vertex, exec->vtx.copied.buffer, + exec->vtx.vertex_size * sizeof(GLfloat)); +} + +static void GLAPIENTRY vbo_exec_EvalCoord1fv( const GLfloat *u ) +{ + vbo_exec_EvalCoord1f( u[0] ); +} + +static void GLAPIENTRY vbo_exec_EvalCoord2fv( const GLfloat *u ) +{ + vbo_exec_EvalCoord2f( u[0], u[1] ); +} + +static void GLAPIENTRY vbo_exec_EvalPoint1( GLint i ) +{ + GET_CURRENT_CONTEXT( ctx ); + GLfloat du = ((ctx->Eval.MapGrid1u2 - ctx->Eval.MapGrid1u1) / + (GLfloat) ctx->Eval.MapGrid1un); + GLfloat u = i * du + ctx->Eval.MapGrid1u1; + + vbo_exec_EvalCoord1f( u ); +} + + +static void GLAPIENTRY vbo_exec_EvalPoint2( GLint i, GLint j ) +{ + GET_CURRENT_CONTEXT( ctx ); + GLfloat du = ((ctx->Eval.MapGrid2u2 - ctx->Eval.MapGrid2u1) / + (GLfloat) ctx->Eval.MapGrid2un); + GLfloat dv = ((ctx->Eval.MapGrid2v2 - ctx->Eval.MapGrid2v1) / + (GLfloat) ctx->Eval.MapGrid2vn); + GLfloat u = i * du + ctx->Eval.MapGrid2u1; + GLfloat v = j * dv + ctx->Eval.MapGrid2v1; + + vbo_exec_EvalCoord2f( u, v ); +} + +/* use noop eval mesh */ +#define vbo_exec_EvalMesh1 _mesa_noop_EvalMesh1 +#define vbo_exec_EvalMesh2 _mesa_noop_EvalMesh2 + +#endif /* FEATURE_evaluators */ + + +/** + * Flush (draw) vertices. + * \param unmap - leave VBO unmapped after flushing? + */ +static void +vbo_exec_FlushVertices_internal(struct vbo_exec_context *exec, GLboolean unmap) +{ + if (exec->vtx.vert_count || unmap) { + vbo_exec_vtx_flush( exec, unmap ); + } + + if (exec->vtx.vertex_size) { + vbo_exec_copy_to_current( exec ); + reset_attrfv( exec ); + } +} + + +/** + * Called via glBegin. + */ +static void GLAPIENTRY vbo_exec_Begin( GLenum mode ) +{ + GET_CURRENT_CONTEXT( ctx ); + + if (ctx->Driver.CurrentExecPrimitive == PRIM_OUTSIDE_BEGIN_END) { + struct vbo_exec_context *exec = &vbo_context(ctx)->exec; + int i; + + if (ctx->NewState) { + _mesa_update_state( ctx ); + + CALL_Begin(ctx->Exec, (mode)); + return; + } + + if (!_mesa_valid_to_render(ctx, "glBegin")) { + return; + } + + /* Heuristic: attempt to isolate attributes occuring outside + * begin/end pairs. + */ + if (exec->vtx.vertex_size && !exec->vtx.attrsz[0]) + vbo_exec_FlushVertices_internal(exec, GL_FALSE); + + i = exec->vtx.prim_count++; + exec->vtx.prim[i].mode = mode; + exec->vtx.prim[i].begin = 1; + exec->vtx.prim[i].end = 0; + exec->vtx.prim[i].indexed = 0; + exec->vtx.prim[i].weak = 0; + exec->vtx.prim[i].pad = 0; + exec->vtx.prim[i].start = exec->vtx.vert_count; + exec->vtx.prim[i].count = 0; + exec->vtx.prim[i].num_instances = 1; + + ctx->Driver.CurrentExecPrimitive = mode; + } + else + _mesa_error( ctx, GL_INVALID_OPERATION, "glBegin" ); + +} + + +/** + * Called via glEnd. + */ +static void GLAPIENTRY vbo_exec_End( void ) +{ + GET_CURRENT_CONTEXT( ctx ); + + if (ctx->Driver.CurrentExecPrimitive != PRIM_OUTSIDE_BEGIN_END) { + struct vbo_exec_context *exec = &vbo_context(ctx)->exec; + int idx = exec->vtx.vert_count; + int i = exec->vtx.prim_count - 1; + + exec->vtx.prim[i].end = 1; + exec->vtx.prim[i].count = idx - exec->vtx.prim[i].start; + + ctx->Driver.CurrentExecPrimitive = PRIM_OUTSIDE_BEGIN_END; + + if (exec->vtx.prim_count == VBO_MAX_PRIM) + vbo_exec_vtx_flush( exec, GL_FALSE ); + } + else + _mesa_error( ctx, GL_INVALID_OPERATION, "glEnd" ); +} + + +/** + * Called via glPrimitiveRestartNV() + */ +static void GLAPIENTRY +vbo_exec_PrimitiveRestartNV(void) +{ + GLenum curPrim; + GET_CURRENT_CONTEXT( ctx ); + + curPrim = ctx->Driver.CurrentExecPrimitive; + + if (curPrim == PRIM_OUTSIDE_BEGIN_END) { + _mesa_error( ctx, GL_INVALID_OPERATION, "glPrimitiveRestartNV" ); + } + else { + vbo_exec_End(); + vbo_exec_Begin(curPrim); + } +} + + + +static void vbo_exec_vtxfmt_init( struct vbo_exec_context *exec ) +{ + GLvertexformat *vfmt = &exec->vtxfmt; + + _MESA_INIT_ARRAYELT_VTXFMT(vfmt, _ae_); + + vfmt->Begin = vbo_exec_Begin; + vfmt->End = vbo_exec_End; + vfmt->PrimitiveRestartNV = vbo_exec_PrimitiveRestartNV; + + _MESA_INIT_DLIST_VTXFMT(vfmt, _mesa_); + _MESA_INIT_EVAL_VTXFMT(vfmt, vbo_exec_); + + vfmt->Rectf = _mesa_noop_Rectf; + + /* from attrib_tmp.h: + */ + vfmt->Color3f = vbo_Color3f; + vfmt->Color3fv = vbo_Color3fv; + vfmt->Color4f = vbo_Color4f; + vfmt->Color4fv = vbo_Color4fv; + vfmt->FogCoordfEXT = vbo_FogCoordfEXT; + vfmt->FogCoordfvEXT = vbo_FogCoordfvEXT; + vfmt->MultiTexCoord1fARB = vbo_MultiTexCoord1f; + vfmt->MultiTexCoord1fvARB = vbo_MultiTexCoord1fv; + vfmt->MultiTexCoord2fARB = vbo_MultiTexCoord2f; + vfmt->MultiTexCoord2fvARB = vbo_MultiTexCoord2fv; + vfmt->MultiTexCoord3fARB = vbo_MultiTexCoord3f; + vfmt->MultiTexCoord3fvARB = vbo_MultiTexCoord3fv; + vfmt->MultiTexCoord4fARB = vbo_MultiTexCoord4f; + vfmt->MultiTexCoord4fvARB = vbo_MultiTexCoord4fv; + vfmt->Normal3f = vbo_Normal3f; + vfmt->Normal3fv = vbo_Normal3fv; + vfmt->SecondaryColor3fEXT = vbo_SecondaryColor3fEXT; + vfmt->SecondaryColor3fvEXT = vbo_SecondaryColor3fvEXT; + vfmt->TexCoord1f = vbo_TexCoord1f; + vfmt->TexCoord1fv = vbo_TexCoord1fv; + vfmt->TexCoord2f = vbo_TexCoord2f; + vfmt->TexCoord2fv = vbo_TexCoord2fv; + vfmt->TexCoord3f = vbo_TexCoord3f; + vfmt->TexCoord3fv = vbo_TexCoord3fv; + vfmt->TexCoord4f = vbo_TexCoord4f; + vfmt->TexCoord4fv = vbo_TexCoord4fv; + vfmt->Vertex2f = vbo_Vertex2f; + vfmt->Vertex2fv = vbo_Vertex2fv; + vfmt->Vertex3f = vbo_Vertex3f; + vfmt->Vertex3fv = vbo_Vertex3fv; + vfmt->Vertex4f = vbo_Vertex4f; + vfmt->Vertex4fv = vbo_Vertex4fv; + + vfmt->VertexAttrib1fARB = vbo_VertexAttrib1fARB; + vfmt->VertexAttrib1fvARB = vbo_VertexAttrib1fvARB; + vfmt->VertexAttrib2fARB = vbo_VertexAttrib2fARB; + vfmt->VertexAttrib2fvARB = vbo_VertexAttrib2fvARB; + vfmt->VertexAttrib3fARB = vbo_VertexAttrib3fARB; + vfmt->VertexAttrib3fvARB = vbo_VertexAttrib3fvARB; + vfmt->VertexAttrib4fARB = vbo_VertexAttrib4fARB; + vfmt->VertexAttrib4fvARB = vbo_VertexAttrib4fvARB; + + vfmt->VertexAttrib1fNV = vbo_VertexAttrib1fNV; + vfmt->VertexAttrib1fvNV = vbo_VertexAttrib1fvNV; + vfmt->VertexAttrib2fNV = vbo_VertexAttrib2fNV; + vfmt->VertexAttrib2fvNV = vbo_VertexAttrib2fvNV; + vfmt->VertexAttrib3fNV = vbo_VertexAttrib3fNV; + vfmt->VertexAttrib3fvNV = vbo_VertexAttrib3fvNV; + vfmt->VertexAttrib4fNV = vbo_VertexAttrib4fNV; + vfmt->VertexAttrib4fvNV = vbo_VertexAttrib4fvNV; + + /* integer-valued */ + vfmt->VertexAttribI1i = vbo_VertexAttribI1i; + vfmt->VertexAttribI2i = vbo_VertexAttribI2i; + vfmt->VertexAttribI3i = vbo_VertexAttribI3i; + vfmt->VertexAttribI4i = vbo_VertexAttribI4i; + vfmt->VertexAttribI2iv = vbo_VertexAttribI2iv; + vfmt->VertexAttribI3iv = vbo_VertexAttribI3iv; + vfmt->VertexAttribI4iv = vbo_VertexAttribI4iv; + + /* unsigned integer-valued */ + vfmt->VertexAttribI1ui = vbo_VertexAttribI1ui; + vfmt->VertexAttribI2ui = vbo_VertexAttribI2ui; + vfmt->VertexAttribI3ui = vbo_VertexAttribI3ui; + vfmt->VertexAttribI4ui = vbo_VertexAttribI4ui; + vfmt->VertexAttribI2uiv = vbo_VertexAttribI2uiv; + vfmt->VertexAttribI3uiv = vbo_VertexAttribI3uiv; + vfmt->VertexAttribI4uiv = vbo_VertexAttribI4uiv; + + vfmt->Materialfv = vbo_Materialfv; + + vfmt->EdgeFlag = vbo_EdgeFlag; + vfmt->Indexf = vbo_Indexf; + vfmt->Indexfv = vbo_Indexfv; + +} + + +#else /* FEATURE_beginend */ + + +static void vbo_exec_vtxfmt_init( struct vbo_exec_context *exec ) +{ + /* silence warnings */ + (void) vbo_Color3f; + (void) vbo_Color3fv; + (void) vbo_Color4f; + (void) vbo_Color4fv; + (void) vbo_FogCoordfEXT; + (void) vbo_FogCoordfvEXT; + (void) vbo_MultiTexCoord1f; + (void) vbo_MultiTexCoord1fv; + (void) vbo_MultiTexCoord2f; + (void) vbo_MultiTexCoord2fv; + (void) vbo_MultiTexCoord3f; + (void) vbo_MultiTexCoord3fv; + (void) vbo_MultiTexCoord4f; + (void) vbo_MultiTexCoord4fv; + (void) vbo_Normal3f; + (void) vbo_Normal3fv; + (void) vbo_SecondaryColor3fEXT; + (void) vbo_SecondaryColor3fvEXT; + (void) vbo_TexCoord1f; + (void) vbo_TexCoord1fv; + (void) vbo_TexCoord2f; + (void) vbo_TexCoord2fv; + (void) vbo_TexCoord3f; + (void) vbo_TexCoord3fv; + (void) vbo_TexCoord4f; + (void) vbo_TexCoord4fv; + (void) vbo_Vertex2f; + (void) vbo_Vertex2fv; + (void) vbo_Vertex3f; + (void) vbo_Vertex3fv; + (void) vbo_Vertex4f; + (void) vbo_Vertex4fv; + + (void) vbo_VertexAttrib1fARB; + (void) vbo_VertexAttrib1fvARB; + (void) vbo_VertexAttrib2fARB; + (void) vbo_VertexAttrib2fvARB; + (void) vbo_VertexAttrib3fARB; + (void) vbo_VertexAttrib3fvARB; + (void) vbo_VertexAttrib4fARB; + (void) vbo_VertexAttrib4fvARB; + + (void) vbo_VertexAttrib1fNV; + (void) vbo_VertexAttrib1fvNV; + (void) vbo_VertexAttrib2fNV; + (void) vbo_VertexAttrib2fvNV; + (void) vbo_VertexAttrib3fNV; + (void) vbo_VertexAttrib3fvNV; + (void) vbo_VertexAttrib4fNV; + (void) vbo_VertexAttrib4fvNV; + + (void) vbo_Materialfv; + + (void) vbo_EdgeFlag; + (void) vbo_Indexf; + (void) vbo_Indexfv; +} + + +#endif /* FEATURE_beginend */ + + +/** + * Tell the VBO module to use a real OpenGL vertex buffer object to + * store accumulated immediate-mode vertex data. + * This replaces the malloced buffer which was created in + * vb_exec_vtx_init() below. + */ +void vbo_use_buffer_objects(struct gl_context *ctx) +{ + struct vbo_exec_context *exec = &vbo_context(ctx)->exec; + /* Any buffer name but 0 can be used here since this bufferobj won't + * go into the bufferobj hashtable. + */ + GLuint bufName = IMM_BUFFER_NAME; + GLenum target = GL_ARRAY_BUFFER_ARB; + GLenum usage = GL_STREAM_DRAW_ARB; + GLsizei size = VBO_VERT_BUFFER_SIZE; + + /* Make sure this func is only used once */ + assert(exec->vtx.bufferobj == ctx->Shared->NullBufferObj); + if (exec->vtx.buffer_map) { + _mesa_align_free(exec->vtx.buffer_map); + exec->vtx.buffer_map = NULL; + exec->vtx.buffer_ptr = NULL; + } + + /* Allocate a real buffer object now */ + _mesa_reference_buffer_object(ctx, &exec->vtx.bufferobj, NULL); + exec->vtx.bufferobj = ctx->Driver.NewBufferObject(ctx, bufName, target); + ctx->Driver.BufferData(ctx, target, size, NULL, usage, exec->vtx.bufferobj); +} + + + +void vbo_exec_vtx_init( struct vbo_exec_context *exec ) +{ + struct gl_context *ctx = exec->ctx; + struct vbo_context *vbo = vbo_context(ctx); + GLuint i; + + /* Allocate a buffer object. Will just reuse this object + * continuously, unless vbo_use_buffer_objects() is called to enable + * use of real VBOs. + */ + _mesa_reference_buffer_object(ctx, + &exec->vtx.bufferobj, + ctx->Shared->NullBufferObj); + + ASSERT(!exec->vtx.buffer_map); + exec->vtx.buffer_map = (GLfloat *)_mesa_align_malloc(VBO_VERT_BUFFER_SIZE, 64); + exec->vtx.buffer_ptr = exec->vtx.buffer_map; + + vbo_exec_vtxfmt_init( exec ); + + /* Hook our functions into the dispatch table. + */ + _mesa_install_exec_vtxfmt( exec->ctx, &exec->vtxfmt ); + + for (i = 0 ; i < VBO_ATTRIB_MAX ; i++) { + ASSERT(i < Elements(exec->vtx.attrsz)); + exec->vtx.attrsz[i] = 0; + ASSERT(i < Elements(exec->vtx.active_sz)); + exec->vtx.active_sz[i] = 0; + } + for (i = 0 ; i < VERT_ATTRIB_MAX; i++) { + ASSERT(i < Elements(exec->vtx.inputs)); + ASSERT(i < Elements(exec->vtx.arrays)); + exec->vtx.inputs[i] = &exec->vtx.arrays[i]; + } + + { + struct gl_client_array *arrays = exec->vtx.arrays; + unsigned i; + + memcpy(arrays, vbo->legacy_currval, 16 * sizeof(arrays[0])); + memcpy(arrays + 16, vbo->generic_currval, 16 * sizeof(arrays[0])); + + for (i = 0; i < 16; ++i) { + arrays[i ].BufferObj = NULL; + arrays[i + 16].BufferObj = NULL; + _mesa_reference_buffer_object(ctx, &arrays[i ].BufferObj, + vbo->legacy_currval[i].BufferObj); + _mesa_reference_buffer_object(ctx, &arrays[i + 16].BufferObj, + vbo->generic_currval[i].BufferObj); + } + } + + exec->vtx.vertex_size = 0; +} + + +void vbo_exec_vtx_destroy( struct vbo_exec_context *exec ) +{ + /* using a real VBO for vertex data */ + struct gl_context *ctx = exec->ctx; + unsigned i; + + /* True VBOs should already be unmapped + */ + if (exec->vtx.buffer_map) { + ASSERT(exec->vtx.bufferobj->Name == 0 || + exec->vtx.bufferobj->Name == IMM_BUFFER_NAME); + if (exec->vtx.bufferobj->Name == 0) { + _mesa_align_free(exec->vtx.buffer_map); + exec->vtx.buffer_map = NULL; + exec->vtx.buffer_ptr = NULL; + } + } + + /* Drop any outstanding reference to the vertex buffer + */ + for (i = 0; i < Elements(exec->vtx.arrays); i++) { + _mesa_reference_buffer_object(ctx, + &exec->vtx.arrays[i].BufferObj, + NULL); + } + + /* Free the vertex buffer. Unmap first if needed. + */ + if (_mesa_bufferobj_mapped(exec->vtx.bufferobj)) { + ctx->Driver.UnmapBuffer(ctx, GL_ARRAY_BUFFER, exec->vtx.bufferobj); + } + _mesa_reference_buffer_object(ctx, &exec->vtx.bufferobj, NULL); +} + + +/** + * Called upon first glVertex, glColor, glTexCoord, etc. + */ +void vbo_exec_BeginVertices( struct gl_context *ctx ) +{ + struct vbo_exec_context *exec = &vbo_context(ctx)->exec; + + vbo_exec_vtx_map( exec ); + + assert((exec->ctx->Driver.NeedFlush & FLUSH_UPDATE_CURRENT) == 0); + exec->ctx->Driver.NeedFlush |= FLUSH_UPDATE_CURRENT; +} + + +/** + * Called via ctx->Driver.FlushVertices() + * \param flags bitmask of FLUSH_STORED_VERTICES, FLUSH_UPDATE_CURRENT + */ +void vbo_exec_FlushVertices( struct gl_context *ctx, GLuint flags ) +{ + struct vbo_exec_context *exec = &vbo_context(ctx)->exec; + +#ifdef DEBUG + /* debug check: make sure we don't get called recursively */ + exec->flush_call_depth++; + assert(exec->flush_call_depth == 1); +#endif + + if (exec->ctx->Driver.CurrentExecPrimitive != PRIM_OUTSIDE_BEGIN_END) { + /* We've had glBegin but not glEnd! */ +#ifdef DEBUG + exec->flush_call_depth--; + assert(exec->flush_call_depth == 0); +#endif + return; + } + + /* Flush (draw), and make sure VBO is left unmapped when done */ + vbo_exec_FlushVertices_internal(exec, GL_TRUE); + + /* Need to do this to ensure BeginVertices gets called again: + */ + if (exec->ctx->Driver.NeedFlush & FLUSH_UPDATE_CURRENT) + exec->ctx->Driver.NeedFlush &= ~FLUSH_UPDATE_CURRENT; + + exec->ctx->Driver.NeedFlush &= ~flags; + +#ifdef DEBUG + exec->flush_call_depth--; + assert(exec->flush_call_depth == 0); +#endif +} + + +static void reset_attrfv( struct vbo_exec_context *exec ) +{ + GLuint i; + + for (i = 0 ; i < VBO_ATTRIB_MAX ; i++) { + exec->vtx.attrsz[i] = 0; + exec->vtx.active_sz[i] = 0; + } + + exec->vtx.vertex_size = 0; +} + + +void GLAPIENTRY +_es_Color4f(GLfloat r, GLfloat g, GLfloat b, GLfloat a) +{ + vbo_Color4f(r, g, b, a); +} + + +void GLAPIENTRY +_es_Normal3f(GLfloat x, GLfloat y, GLfloat z) +{ + vbo_Normal3f(x, y, z); +} + + +void GLAPIENTRY +_es_MultiTexCoord4f(GLenum target, GLfloat s, GLfloat t, GLfloat r, GLfloat q) +{ + vbo_MultiTexCoord4f(target, s, t, r, q); +} + + +void GLAPIENTRY +_es_Materialfv(GLenum face, GLenum pname, const GLfloat *params) +{ + vbo_Materialfv(face, pname, params); +} + + +void GLAPIENTRY +_es_Materialf(GLenum face, GLenum pname, GLfloat param) +{ + GLfloat p[4]; + p[0] = param; + p[1] = p[2] = p[3] = 0.0F; + vbo_Materialfv(face, pname, p); +} + + +/** + * A special version of glVertexAttrib4f that does not treat index 0 as + * VBO_ATTRIB_POS. + */ +static void +VertexAttrib4f_nopos(GLuint index, GLfloat x, GLfloat y, GLfloat z, GLfloat w) +{ + GET_CURRENT_CONTEXT(ctx); + if (index < MAX_VERTEX_GENERIC_ATTRIBS) + ATTR(VBO_ATTRIB_GENERIC0 + index, 4, x, y, z, w); + else + ERROR(); +} + +void GLAPIENTRY +_es_VertexAttrib4f(GLuint index, GLfloat x, GLfloat y, GLfloat z, GLfloat w) +{ + VertexAttrib4f_nopos(index, x, y, z, w); +} + + +void GLAPIENTRY +_es_VertexAttrib1f(GLuint indx, GLfloat x) +{ + VertexAttrib4f_nopos(indx, x, 0.0f, 0.0f, 1.0f); +} + + +void GLAPIENTRY +_es_VertexAttrib1fv(GLuint indx, const GLfloat* values) +{ + VertexAttrib4f_nopos(indx, values[0], 0.0f, 0.0f, 1.0f); +} + + +void GLAPIENTRY +_es_VertexAttrib2f(GLuint indx, GLfloat x, GLfloat y) +{ + VertexAttrib4f_nopos(indx, x, y, 0.0f, 1.0f); +} + + +void GLAPIENTRY +_es_VertexAttrib2fv(GLuint indx, const GLfloat* values) +{ + VertexAttrib4f_nopos(indx, values[0], values[1], 0.0f, 1.0f); +} + + +void GLAPIENTRY +_es_VertexAttrib3f(GLuint indx, GLfloat x, GLfloat y, GLfloat z) +{ + VertexAttrib4f_nopos(indx, x, y, z, 1.0f); +} + + +void GLAPIENTRY +_es_VertexAttrib3fv(GLuint indx, const GLfloat* values) +{ + VertexAttrib4f_nopos(indx, values[0], values[1], values[2], 1.0f); +} + + +void GLAPIENTRY +_es_VertexAttrib4fv(GLuint indx, const GLfloat* values) +{ + VertexAttrib4f_nopos(indx, values[0], values[1], values[2], values[3]); +} diff --git a/mesalib/src/mesa/vbo/vbo_exec_draw.c b/mesalib/src/mesa/vbo/vbo_exec_draw.c index f8be83ea8..539658021 100644 --- a/mesalib/src/mesa/vbo/vbo_exec_draw.c +++ b/mesalib/src/mesa/vbo/vbo_exec_draw.c @@ -253,6 +253,9 @@ vbo_exec_bind_arrays( struct gl_context *ctx ) } +/** + * Unmap the VBO. This is called before drawing. + */ static void vbo_exec_vtx_unmap( struct vbo_exec_context *exec ) { @@ -285,6 +288,9 @@ vbo_exec_vtx_unmap( struct vbo_exec_context *exec ) } +/** + * Map the vertex buffer to begin storing glVertex, glColor, etc data. + */ void vbo_exec_vtx_map( struct vbo_exec_context *exec ) { @@ -301,14 +307,12 @@ vbo_exec_vtx_map( struct vbo_exec_context *exec ) if (!_mesa_is_bufferobj(exec->vtx.bufferobj)) return; - if (exec->vtx.buffer_map != NULL) { - assert(0); - exec->vtx.buffer_map = NULL; - exec->vtx.buffer_ptr = NULL; - } + assert(!exec->vtx.buffer_map); + assert(!exec->vtx.buffer_ptr); if (VBO_VERT_BUFFER_SIZE > exec->vtx.buffer_used + 1024 && ctx->Driver.MapBufferRange) { + /* The VBO exists and there's room for more */ exec->vtx.buffer_map = (GLfloat *)ctx->Driver.MapBufferRange(ctx, target, @@ -321,6 +325,7 @@ vbo_exec_vtx_map( struct vbo_exec_context *exec ) } if (!exec->vtx.buffer_map) { + /* Need to allocate a new VBO */ exec->vtx.buffer_used = 0; ctx->Driver.BufferData(ctx, target, @@ -349,9 +354,10 @@ vbo_exec_vtx_map( struct vbo_exec_context *exec ) /** * Execute the buffer and save copied verts. + * \param keep_unmapped if true, leave the VBO unmapped when we're done. */ void -vbo_exec_vtx_flush( struct vbo_exec_context *exec, GLboolean unmap ) +vbo_exec_vtx_flush(struct vbo_exec_context *exec, GLboolean keepUnmapped) { if (0) vbo_exec_debug_verts( exec ); @@ -391,7 +397,7 @@ vbo_exec_vtx_flush( struct vbo_exec_context *exec, GLboolean unmap ) /* If using a real VBO, get new storage -- unless asked not to. */ - if (_mesa_is_bufferobj(exec->vtx.bufferobj) && !unmap) { + if (_mesa_is_bufferobj(exec->vtx.bufferobj) && !keepUnmapped) { vbo_exec_vtx_map( exec ); } } @@ -399,14 +405,13 @@ vbo_exec_vtx_flush( struct vbo_exec_context *exec, GLboolean unmap ) /* May have to unmap explicitly if we didn't draw: */ - if (unmap && + if (keepUnmapped && _mesa_is_bufferobj(exec->vtx.bufferobj) && exec->vtx.buffer_map) { vbo_exec_vtx_unmap( exec ); } - - if (unmap || exec->vtx.vertex_size == 0) + if (keepUnmapped || exec->vtx.vertex_size == 0) exec->vtx.max_vert = 0; else exec->vtx.max_vert = ((VBO_VERT_BUFFER_SIZE - exec->vtx.buffer_used) / diff --git a/pixman/configure.ac b/pixman/configure.ac index 5242799bb..8d96647f9 100644 --- a/pixman/configure.ac +++ b/pixman/configure.ac @@ -326,7 +326,7 @@ if test "x$SSE2_CFLAGS" = "x" ; then SSE2_CFLAGS="-xarch=sse2" fi else - SSE2_CFLAGS="-mmmx -msse2 -Winline" + SSE2_CFLAGS="-msse2 -Winline" fi fi diff --git a/pixman/pixman/pixman-sse2.c b/pixman/pixman/pixman-sse2.c index 2e135e2fe..88287b453 100644 --- a/pixman/pixman/pixman-sse2.c +++ b/pixman/pixman/pixman-sse2.c @@ -30,36 +30,12 @@ #include <config.h> #endif -#include <mmintrin.h> #include <xmmintrin.h> /* for _mm_shuffle_pi16 and _MM_SHUFFLE */ #include <emmintrin.h> /* for SSE2 intrinsics */ #include "pixman-private.h" #include "pixman-combine32.h" #include "pixman-fast-path.h" -#if defined(_MSC_VER) && defined(_M_AMD64) -/* Windows 64 doesn't allow MMX to be used, so - * the pixman-x64-mmx-emulation.h file contains - * implementations of those MMX intrinsics that - * are used in the SSE2 implementation. - */ -# include "pixman-x64-mmx-emulation.h" -#endif - -#ifdef USE_SSE2 - -/* -------------------------------------------------------------------- - * Locals - */ - -static __m64 mask_x0080; -static __m64 mask_x00ff; -static __m64 mask_x0101; -static __m64 mask_x_alpha; - -static __m64 mask_x565_rgb; -static __m64 mask_x565_unpack; - static __m128i mask_0080; static __m128i mask_00ff; static __m128i mask_0101; @@ -77,9 +53,6 @@ static __m128i mask_blue; static __m128i mask_565_fix_rb; static __m128i mask_565_fix_g; -/* ---------------------------------------------------------------------- - * SSE2 Inlines - */ static force_inline __m128i unpack_32_1x128 (uint32_t data) { @@ -397,146 +370,104 @@ save_128_unaligned (__m128i* dst, _mm_storeu_si128 (dst, data); } -/* ------------------------------------------------------------------ - * MMX inlines - */ - -static force_inline __m64 -load_32_1x64 (uint32_t data) -{ - return _mm_cvtsi32_si64 (data); -} - -static force_inline __m64 -unpack_32_1x64 (uint32_t data) -{ - return _mm_unpacklo_pi8 (load_32_1x64 (data), _mm_setzero_si64 ()); -} - -static force_inline __m64 -expand_alpha_1x64 (__m64 data) +static force_inline __m128i +load_32_1x128 (uint32_t data) { - return _mm_shuffle_pi16 (data, _MM_SHUFFLE (3, 3, 3, 3)); + return _mm_cvtsi32_si128 (data); } -static force_inline __m64 -expand_alpha_rev_1x64 (__m64 data) +static force_inline __m128i +expand_alpha_rev_1x128 (__m128i data) { - return _mm_shuffle_pi16 (data, _MM_SHUFFLE (0, 0, 0, 0)); + return _mm_shufflelo_epi16 (data, _MM_SHUFFLE (0, 0, 0, 0)); } -static force_inline __m64 -expand_pixel_8_1x64 (uint8_t data) +static force_inline __m128i +expand_pixel_8_1x128 (uint8_t data) { - return _mm_shuffle_pi16 ( - unpack_32_1x64 ((uint32_t)data), _MM_SHUFFLE (0, 0, 0, 0)); + return _mm_shufflelo_epi16 ( + unpack_32_1x128 ((uint32_t)data), _MM_SHUFFLE (0, 0, 0, 0)); } -static force_inline __m64 -pix_multiply_1x64 (__m64 data, - __m64 alpha) +static force_inline __m128i +pix_multiply_1x128 (__m128i data, + __m128i alpha) { - return _mm_mulhi_pu16 (_mm_adds_pu16 (_mm_mullo_pi16 (data, alpha), - mask_x0080), - mask_x0101); + return _mm_mulhi_epu16 (_mm_adds_epu16 (_mm_mullo_epi16 (data, alpha), + mask_0080), + mask_0101); } -static force_inline __m64 -pix_add_multiply_1x64 (__m64* src, - __m64* alpha_dst, - __m64* dst, - __m64* alpha_src) +static force_inline __m128i +pix_add_multiply_1x128 (__m128i* src, + __m128i* alpha_dst, + __m128i* dst, + __m128i* alpha_src) { - __m64 t1 = pix_multiply_1x64 (*src, *alpha_dst); - __m64 t2 = pix_multiply_1x64 (*dst, *alpha_src); + __m128i t1 = pix_multiply_1x128 (*src, *alpha_dst); + __m128i t2 = pix_multiply_1x128 (*dst, *alpha_src); - return _mm_adds_pu8 (t1, t2); + return _mm_adds_epu8 (t1, t2); } -static force_inline __m64 -negate_1x64 (__m64 data) +static force_inline __m128i +negate_1x128 (__m128i data) { - return _mm_xor_si64 (data, mask_x00ff); + return _mm_xor_si128 (data, mask_00ff); } -static force_inline __m64 -invert_colors_1x64 (__m64 data) +static force_inline __m128i +invert_colors_1x128 (__m128i data) { - return _mm_shuffle_pi16 (data, _MM_SHUFFLE (3, 0, 1, 2)); + return _mm_shufflelo_epi16 (data, _MM_SHUFFLE (3, 0, 1, 2)); } -static force_inline __m64 -over_1x64 (__m64 src, __m64 alpha, __m64 dst) +static force_inline __m128i +over_1x128 (__m128i src, __m128i alpha, __m128i dst) { - return _mm_adds_pu8 (src, pix_multiply_1x64 (dst, negate_1x64 (alpha))); + return _mm_adds_epu8 (src, pix_multiply_1x128 (dst, negate_1x128 (alpha))); } -static force_inline __m64 -in_over_1x64 (__m64* src, __m64* alpha, __m64* mask, __m64* dst) +static force_inline __m128i +in_over_1x128 (__m128i* src, __m128i* alpha, __m128i* mask, __m128i* dst) { - return over_1x64 (pix_multiply_1x64 (*src, *mask), - pix_multiply_1x64 (*alpha, *mask), - *dst); + return over_1x128 (pix_multiply_1x128 (*src, *mask), + pix_multiply_1x128 (*alpha, *mask), + *dst); } -static force_inline __m64 -over_rev_non_pre_1x64 (__m64 src, __m64 dst) +static force_inline __m128i +over_rev_non_pre_1x128 (__m128i src, __m128i dst) { - __m64 alpha = expand_alpha_1x64 (src); + __m128i alpha = expand_alpha_1x128 (src); - return over_1x64 (pix_multiply_1x64 (invert_colors_1x64 (src), - _mm_or_si64 (alpha, mask_x_alpha)), - alpha, - dst); + return over_1x128 (pix_multiply_1x128 (invert_colors_1x128 (src), + _mm_or_si128 (alpha, mask_alpha)), + alpha, + dst); } static force_inline uint32_t -pack_1x64_32 (__m64 data) +pack_1x128_32 (__m128i data) { - return _mm_cvtsi64_si32 (_mm_packs_pu16 (data, _mm_setzero_si64 ())); + return _mm_cvtsi128_si32 (_mm_packus_epi16 (data, _mm_setzero_si128 ())); } -/* Expand 16 bits positioned at @pos (0-3) of a mmx register into - * - * 00RR00GG00BB - * - * --- Expanding 565 in the low word --- - * - * m = (m << (32 - 3)) | (m << (16 - 5)) | m; - * m = m & (01f0003f001f); - * m = m * (008404100840); - * m = m >> 8; - * - * Note the trick here - the top word is shifted by another nibble to - * avoid it bumping into the middle word - */ -static force_inline __m64 -expand565_16_1x64 (uint16_t pixel) +static force_inline __m128i +expand565_16_1x128 (uint16_t pixel) { - __m64 p; - __m64 t1, t2; + __m128i m = _mm_cvtsi32_si128 (pixel); - p = _mm_cvtsi32_si64 ((uint32_t) pixel); + m = unpack_565_to_8888 (m); - t1 = _mm_slli_si64 (p, 36 - 11); - t2 = _mm_slli_si64 (p, 16 - 5); - - p = _mm_or_si64 (t1, p); - p = _mm_or_si64 (t2, p); - p = _mm_and_si64 (p, mask_x565_rgb); - p = _mm_mullo_pi16 (p, mask_x565_unpack); - - return _mm_srli_pi16 (p, 8); + return _mm_unpacklo_epi8 (m, _mm_setzero_si128 ()); } -/* ---------------------------------------------------------------------------- - * Compose Core transformations - */ static force_inline uint32_t core_combine_over_u_pixel_sse2 (uint32_t src, uint32_t dst) { uint8_t a; - __m64 ms; + __m128i xmms; a = src >> 24; @@ -546,9 +477,10 @@ core_combine_over_u_pixel_sse2 (uint32_t src, uint32_t dst) } else if (src) { - ms = unpack_32_1x64 (src); - return pack_1x64_32 ( - over_1x64 (ms, expand_alpha_1x64 (ms), unpack_32_1x64 (dst))); + xmms = unpack_32_1x128 (src); + return pack_1x128_32 ( + over_1x128 (xmms, expand_alpha_1x128 (xmms), + unpack_32_1x128 (dst))); } return dst; @@ -561,15 +493,15 @@ combine1 (const uint32_t *ps, const uint32_t *pm) if (pm) { - __m64 ms, mm; + __m128i ms, mm; - mm = unpack_32_1x64 (*pm); - mm = expand_alpha_1x64 (mm); + mm = unpack_32_1x128 (*pm); + mm = expand_alpha_1x128 (mm); - ms = unpack_32_1x64 (s); - ms = pix_multiply_1x64 (ms, mm); + ms = unpack_32_1x128 (s); + ms = pix_multiply_1x128 (ms, mm); - s = pack_1x64_32 (ms); + s = pack_1x128_32 (ms); } return s; @@ -766,10 +698,12 @@ core_combine_over_u_sse2_no_mask (uint32_t * pd, } static force_inline void -core_combine_over_u_sse2 (uint32_t* pd, - const uint32_t* ps, - const uint32_t* pm, - int w) +sse2_combine_over_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * pd, + const uint32_t * ps, + const uint32_t * pm, + int w) { if (pm) core_combine_over_u_sse2_mask (pd, ps, pm, w); @@ -777,11 +711,13 @@ core_combine_over_u_sse2 (uint32_t* pd, core_combine_over_u_sse2_no_mask (pd, ps, w); } -static force_inline void -core_combine_over_reverse_u_sse2 (uint32_t* pd, - const uint32_t* ps, - const uint32_t* pm, - int w) +static void +sse2_combine_over_reverse_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * pd, + const uint32_t * ps, + const uint32_t * pm, + int w) { uint32_t s, d; @@ -847,7 +783,7 @@ core_combine_over_reverse_u_sse2 (uint32_t* pd, } static force_inline uint32_t -core_combine_in_u_pixelsse2 (uint32_t src, uint32_t dst) +core_combine_in_u_pixel_sse2 (uint32_t src, uint32_t dst) { uint32_t maska = src >> 24; @@ -857,19 +793,21 @@ core_combine_in_u_pixelsse2 (uint32_t src, uint32_t dst) } else if (maska != 0xff) { - return pack_1x64_32 ( - pix_multiply_1x64 (unpack_32_1x64 (dst), - expand_alpha_1x64 (unpack_32_1x64 (src)))); + return pack_1x128_32 ( + pix_multiply_1x128 (unpack_32_1x128 (dst), + expand_alpha_1x128 (unpack_32_1x128 (src)))); } return dst; } -static force_inline void -core_combine_in_u_sse2 (uint32_t* pd, - const uint32_t* ps, - const uint32_t* pm, - int w) +static void +sse2_combine_in_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * pd, + const uint32_t * ps, + const uint32_t * pm, + int w) { uint32_t s, d; @@ -881,7 +819,7 @@ core_combine_in_u_sse2 (uint32_t* pd, s = combine1 (ps, pm); d = *pd; - *pd++ = core_combine_in_u_pixelsse2 (d, s); + *pd++ = core_combine_in_u_pixel_sse2 (d, s); w--; ps++; if (pm) @@ -916,7 +854,7 @@ core_combine_in_u_sse2 (uint32_t* pd, s = combine1 (ps, pm); d = *pd; - *pd++ = core_combine_in_u_pixelsse2 (d, s); + *pd++ = core_combine_in_u_pixel_sse2 (d, s); w--; ps++; if (pm) @@ -924,11 +862,13 @@ core_combine_in_u_sse2 (uint32_t* pd, } } -static force_inline void -core_combine_reverse_in_u_sse2 (uint32_t* pd, - const uint32_t* ps, - const uint32_t *pm, - int w) +static void +sse2_combine_in_reverse_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * pd, + const uint32_t * ps, + const uint32_t * pm, + int w) { uint32_t s, d; @@ -940,7 +880,7 @@ core_combine_reverse_in_u_sse2 (uint32_t* pd, s = combine1 (ps, pm); d = *pd; - *pd++ = core_combine_in_u_pixelsse2 (s, d); + *pd++ = core_combine_in_u_pixel_sse2 (s, d); ps++; w--; if (pm) @@ -975,7 +915,7 @@ core_combine_reverse_in_u_sse2 (uint32_t* pd, s = combine1 (ps, pm); d = *pd; - *pd++ = core_combine_in_u_pixelsse2 (s, d); + *pd++ = core_combine_in_u_pixel_sse2 (s, d); w--; ps++; if (pm) @@ -983,21 +923,23 @@ core_combine_reverse_in_u_sse2 (uint32_t* pd, } } -static force_inline void -core_combine_reverse_out_u_sse2 (uint32_t* pd, - const uint32_t* ps, - const uint32_t* pm, - int w) +static void +sse2_combine_out_reverse_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * pd, + const uint32_t * ps, + const uint32_t * pm, + int w) { while (w && ((unsigned long) pd & 15)) { uint32_t s = combine1 (ps, pm); uint32_t d = *pd; - *pd++ = pack_1x64_32 ( - pix_multiply_1x64 ( - unpack_32_1x64 (d), negate_1x64 ( - expand_alpha_1x64 (unpack_32_1x64 (s))))); + *pd++ = pack_1x128_32 ( + pix_multiply_1x128 ( + unpack_32_1x128 (d), negate_1x128 ( + expand_alpha_1x128 (unpack_32_1x128 (s))))); if (pm) pm++; @@ -1039,10 +981,10 @@ core_combine_reverse_out_u_sse2 (uint32_t* pd, uint32_t s = combine1 (ps, pm); uint32_t d = *pd; - *pd++ = pack_1x64_32 ( - pix_multiply_1x64 ( - unpack_32_1x64 (d), negate_1x64 ( - expand_alpha_1x64 (unpack_32_1x64 (s))))); + *pd++ = pack_1x128_32 ( + pix_multiply_1x128 ( + unpack_32_1x128 (d), negate_1x128 ( + expand_alpha_1x128 (unpack_32_1x128 (s))))); ps++; if (pm) pm++; @@ -1050,21 +992,23 @@ core_combine_reverse_out_u_sse2 (uint32_t* pd, } } -static force_inline void -core_combine_out_u_sse2 (uint32_t* pd, - const uint32_t* ps, - const uint32_t* pm, - int w) +static void +sse2_combine_out_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * pd, + const uint32_t * ps, + const uint32_t * pm, + int w) { while (w && ((unsigned long) pd & 15)) { uint32_t s = combine1 (ps, pm); uint32_t d = *pd; - *pd++ = pack_1x64_32 ( - pix_multiply_1x64 ( - unpack_32_1x64 (s), negate_1x64 ( - expand_alpha_1x64 (unpack_32_1x64 (d))))); + *pd++ = pack_1x128_32 ( + pix_multiply_1x128 ( + unpack_32_1x128 (s), negate_1x128 ( + expand_alpha_1x128 (unpack_32_1x128 (d))))); w--; ps++; if (pm) @@ -1104,10 +1048,10 @@ core_combine_out_u_sse2 (uint32_t* pd, uint32_t s = combine1 (ps, pm); uint32_t d = *pd; - *pd++ = pack_1x64_32 ( - pix_multiply_1x64 ( - unpack_32_1x64 (s), negate_1x64 ( - expand_alpha_1x64 (unpack_32_1x64 (d))))); + *pd++ = pack_1x128_32 ( + pix_multiply_1x128 ( + unpack_32_1x128 (s), negate_1x128 ( + expand_alpha_1x128 (unpack_32_1x128 (d))))); w--; ps++; if (pm) @@ -1119,20 +1063,22 @@ static force_inline uint32_t core_combine_atop_u_pixel_sse2 (uint32_t src, uint32_t dst) { - __m64 s = unpack_32_1x64 (src); - __m64 d = unpack_32_1x64 (dst); + __m128i s = unpack_32_1x128 (src); + __m128i d = unpack_32_1x128 (dst); - __m64 sa = negate_1x64 (expand_alpha_1x64 (s)); - __m64 da = expand_alpha_1x64 (d); + __m128i sa = negate_1x128 (expand_alpha_1x128 (s)); + __m128i da = expand_alpha_1x128 (d); - return pack_1x64_32 (pix_add_multiply_1x64 (&s, &da, &d, &sa)); + return pack_1x128_32 (pix_add_multiply_1x128 (&s, &da, &d, &sa)); } -static force_inline void -core_combine_atop_u_sse2 (uint32_t* pd, - const uint32_t* ps, - const uint32_t* pm, - int w) +static void +sse2_combine_atop_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * pd, + const uint32_t * ps, + const uint32_t * pm, + int w) { uint32_t s, d; @@ -1201,20 +1147,22 @@ static force_inline uint32_t core_combine_reverse_atop_u_pixel_sse2 (uint32_t src, uint32_t dst) { - __m64 s = unpack_32_1x64 (src); - __m64 d = unpack_32_1x64 (dst); + __m128i s = unpack_32_1x128 (src); + __m128i d = unpack_32_1x128 (dst); - __m64 sa = expand_alpha_1x64 (s); - __m64 da = negate_1x64 (expand_alpha_1x64 (d)); + __m128i sa = expand_alpha_1x128 (s); + __m128i da = negate_1x128 (expand_alpha_1x128 (d)); - return pack_1x64_32 (pix_add_multiply_1x64 (&s, &da, &d, &sa)); + return pack_1x128_32 (pix_add_multiply_1x128 (&s, &da, &d, &sa)); } -static force_inline void -core_combine_reverse_atop_u_sse2 (uint32_t* pd, - const uint32_t* ps, - const uint32_t* pm, - int w) +static void +sse2_combine_atop_reverse_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * pd, + const uint32_t * ps, + const uint32_t * pm, + int w) { uint32_t s, d; @@ -1283,20 +1231,22 @@ static force_inline uint32_t core_combine_xor_u_pixel_sse2 (uint32_t src, uint32_t dst) { - __m64 s = unpack_32_1x64 (src); - __m64 d = unpack_32_1x64 (dst); + __m128i s = unpack_32_1x128 (src); + __m128i d = unpack_32_1x128 (dst); - __m64 neg_d = negate_1x64 (expand_alpha_1x64 (d)); - __m64 neg_s = negate_1x64 (expand_alpha_1x64 (s)); + __m128i neg_d = negate_1x128 (expand_alpha_1x128 (d)); + __m128i neg_s = negate_1x128 (expand_alpha_1x128 (s)); - return pack_1x64_32 (pix_add_multiply_1x64 (&s, &neg_d, &d, &neg_s)); + return pack_1x128_32 (pix_add_multiply_1x128 (&s, &neg_d, &d, &neg_s)); } -static force_inline void -core_combine_xor_u_sse2 (uint32_t* dst, - const uint32_t* src, - const uint32_t *mask, - int width) +static void +sse2_combine_xor_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dst, + const uint32_t * src, + const uint32_t * mask, + int width) { int w = width; uint32_t s, d; @@ -1368,10 +1318,12 @@ core_combine_xor_u_sse2 (uint32_t* dst, } static force_inline void -core_combine_add_u_sse2 (uint32_t* dst, - const uint32_t* src, - const uint32_t* mask, - int width) +sse2_combine_add_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dst, + const uint32_t * src, + const uint32_t * mask, + int width) { int w = width; uint32_t s, d; @@ -1387,8 +1339,8 @@ core_combine_add_u_sse2 (uint32_t* dst, ps++; if (pm) pm++; - *pd++ = _mm_cvtsi64_si32 ( - _mm_adds_pu8 (_mm_cvtsi32_si64 (s), _mm_cvtsi32_si64 (d))); + *pd++ = _mm_cvtsi128_si32 ( + _mm_adds_epu8 (_mm_cvtsi32_si128 (s), _mm_cvtsi32_si128 (d))); w--; } @@ -1414,8 +1366,8 @@ core_combine_add_u_sse2 (uint32_t* dst, d = *pd; ps++; - *pd++ = _mm_cvtsi64_si32 ( - _mm_adds_pu8 (_mm_cvtsi32_si64 (s), _mm_cvtsi32_si64 (d))); + *pd++ = _mm_cvtsi128_si32 ( + _mm_adds_epu8 (_mm_cvtsi32_si128 (s), _mm_cvtsi32_si128 (d))); if (pm) pm++; } @@ -1425,25 +1377,27 @@ static force_inline uint32_t core_combine_saturate_u_pixel_sse2 (uint32_t src, uint32_t dst) { - __m64 ms = unpack_32_1x64 (src); - __m64 md = unpack_32_1x64 (dst); + __m128i ms = unpack_32_1x128 (src); + __m128i md = unpack_32_1x128 (dst); uint32_t sa = src >> 24; uint32_t da = ~dst >> 24; if (sa > da) { - ms = pix_multiply_1x64 ( - ms, expand_alpha_1x64 (unpack_32_1x64 (DIV_UN8 (da, sa) << 24))); + ms = pix_multiply_1x128 ( + ms, expand_alpha_1x128 (unpack_32_1x128 (DIV_UN8 (da, sa) << 24))); } - return pack_1x64_32 (_mm_adds_pu16 (md, ms)); + return pack_1x128_32 (_mm_adds_epu16 (md, ms)); } -static force_inline void -core_combine_saturate_u_sse2 (uint32_t * pd, - const uint32_t *ps, - const uint32_t *pm, - int w) +static void +sse2_combine_saturate_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * pd, + const uint32_t * ps, + const uint32_t * pm, + int w) { uint32_t s, d; @@ -1524,11 +1478,13 @@ core_combine_saturate_u_sse2 (uint32_t * pd, } } -static force_inline void -core_combine_src_ca_sse2 (uint32_t* pd, - const uint32_t* ps, - const uint32_t *pm, - int w) +static void +sse2_combine_src_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * pd, + const uint32_t * ps, + const uint32_t * pm, + int w) { uint32_t s, m; @@ -1540,8 +1496,8 @@ core_combine_src_ca_sse2 (uint32_t* pd, { s = *ps++; m = *pm++; - *pd++ = pack_1x64_32 ( - pix_multiply_1x64 (unpack_32_1x64 (s), unpack_32_1x64 (m))); + *pd++ = pack_1x128_32 ( + pix_multiply_1x128 (unpack_32_1x128 (s), unpack_32_1x128 (m))); w--; } @@ -1570,8 +1526,8 @@ core_combine_src_ca_sse2 (uint32_t* pd, { s = *ps++; m = *pm++; - *pd++ = pack_1x64_32 ( - pix_multiply_1x64 (unpack_32_1x64 (s), unpack_32_1x64 (m))); + *pd++ = pack_1x128_32 ( + pix_multiply_1x128 (unpack_32_1x128 (s), unpack_32_1x128 (m))); w--; } } @@ -1581,19 +1537,21 @@ core_combine_over_ca_pixel_sse2 (uint32_t src, uint32_t mask, uint32_t dst) { - __m64 s = unpack_32_1x64 (src); - __m64 expAlpha = expand_alpha_1x64 (s); - __m64 unpk_mask = unpack_32_1x64 (mask); - __m64 unpk_dst = unpack_32_1x64 (dst); + __m128i s = unpack_32_1x128 (src); + __m128i expAlpha = expand_alpha_1x128 (s); + __m128i unpk_mask = unpack_32_1x128 (mask); + __m128i unpk_dst = unpack_32_1x128 (dst); - return pack_1x64_32 (in_over_1x64 (&s, &expAlpha, &unpk_mask, &unpk_dst)); + return pack_1x128_32 (in_over_1x128 (&s, &expAlpha, &unpk_mask, &unpk_dst)); } -static force_inline void -core_combine_over_ca_sse2 (uint32_t* pd, - const uint32_t* ps, - const uint32_t *pm, - int w) +static void +sse2_combine_over_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * pd, + const uint32_t * ps, + const uint32_t * pm, + int w) { uint32_t s, m, d; @@ -1655,19 +1613,21 @@ core_combine_over_reverse_ca_pixel_sse2 (uint32_t src, uint32_t mask, uint32_t dst) { - __m64 d = unpack_32_1x64 (dst); + __m128i d = unpack_32_1x128 (dst); - return pack_1x64_32 ( - over_1x64 (d, expand_alpha_1x64 (d), - pix_multiply_1x64 (unpack_32_1x64 (src), - unpack_32_1x64 (mask)))); + return pack_1x128_32 ( + over_1x128 (d, expand_alpha_1x128 (d), + pix_multiply_1x128 (unpack_32_1x128 (src), + unpack_32_1x128 (mask)))); } -static force_inline void -core_combine_over_reverse_ca_sse2 (uint32_t* pd, - const uint32_t* ps, - const uint32_t *pm, - int w) +static void +sse2_combine_over_reverse_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * pd, + const uint32_t * ps, + const uint32_t * pm, + int w) { uint32_t s, m, d; @@ -1726,11 +1686,13 @@ core_combine_over_reverse_ca_sse2 (uint32_t* pd, } } -static force_inline void -core_combine_in_ca_sse2 (uint32_t * pd, - const uint32_t *ps, - const uint32_t *pm, - int w) +static void +sse2_combine_in_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * pd, + const uint32_t * ps, + const uint32_t * pm, + int w) { uint32_t s, m, d; @@ -1745,10 +1707,10 @@ core_combine_in_ca_sse2 (uint32_t * pd, m = *pm++; d = *pd; - *pd++ = pack_1x64_32 ( - pix_multiply_1x64 ( - pix_multiply_1x64 (unpack_32_1x64 (s), unpack_32_1x64 (m)), - expand_alpha_1x64 (unpack_32_1x64 (d)))); + *pd++ = pack_1x128_32 ( + pix_multiply_1x128 ( + pix_multiply_1x128 (unpack_32_1x128 (s), unpack_32_1x128 (m)), + expand_alpha_1x128 (unpack_32_1x128 (d)))); w--; } @@ -1789,21 +1751,23 @@ core_combine_in_ca_sse2 (uint32_t * pd, m = *pm++; d = *pd; - *pd++ = pack_1x64_32 ( - pix_multiply_1x64 ( - pix_multiply_1x64 ( - unpack_32_1x64 (s), unpack_32_1x64 (m)), - expand_alpha_1x64 (unpack_32_1x64 (d)))); + *pd++ = pack_1x128_32 ( + pix_multiply_1x128 ( + pix_multiply_1x128 ( + unpack_32_1x128 (s), unpack_32_1x128 (m)), + expand_alpha_1x128 (unpack_32_1x128 (d)))); w--; } } -static force_inline void -core_combine_in_reverse_ca_sse2 (uint32_t * pd, - const uint32_t *ps, - const uint32_t *pm, - int w) +static void +sse2_combine_in_reverse_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * pd, + const uint32_t * ps, + const uint32_t * pm, + int w) { uint32_t s, m, d; @@ -1818,11 +1782,11 @@ core_combine_in_reverse_ca_sse2 (uint32_t * pd, m = *pm++; d = *pd; - *pd++ = pack_1x64_32 ( - pix_multiply_1x64 ( - unpack_32_1x64 (d), - pix_multiply_1x64 (unpack_32_1x64 (m), - expand_alpha_1x64 (unpack_32_1x64 (s))))); + *pd++ = pack_1x128_32 ( + pix_multiply_1x128 ( + unpack_32_1x128 (d), + pix_multiply_1x128 (unpack_32_1x128 (m), + expand_alpha_1x128 (unpack_32_1x128 (s))))); w--; } @@ -1861,20 +1825,22 @@ core_combine_in_reverse_ca_sse2 (uint32_t * pd, m = *pm++; d = *pd; - *pd++ = pack_1x64_32 ( - pix_multiply_1x64 ( - unpack_32_1x64 (d), - pix_multiply_1x64 (unpack_32_1x64 (m), - expand_alpha_1x64 (unpack_32_1x64 (s))))); + *pd++ = pack_1x128_32 ( + pix_multiply_1x128 ( + unpack_32_1x128 (d), + pix_multiply_1x128 (unpack_32_1x128 (m), + expand_alpha_1x128 (unpack_32_1x128 (s))))); w--; } } -static force_inline void -core_combine_out_ca_sse2 (uint32_t * pd, - const uint32_t *ps, - const uint32_t *pm, - int w) +static void +sse2_combine_out_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * pd, + const uint32_t * ps, + const uint32_t * pm, + int w) { uint32_t s, m, d; @@ -1889,11 +1855,11 @@ core_combine_out_ca_sse2 (uint32_t * pd, m = *pm++; d = *pd; - *pd++ = pack_1x64_32 ( - pix_multiply_1x64 ( - pix_multiply_1x64 ( - unpack_32_1x64 (s), unpack_32_1x64 (m)), - negate_1x64 (expand_alpha_1x64 (unpack_32_1x64 (d))))); + *pd++ = pack_1x128_32 ( + pix_multiply_1x128 ( + pix_multiply_1x128 ( + unpack_32_1x128 (s), unpack_32_1x128 (m)), + negate_1x128 (expand_alpha_1x128 (unpack_32_1x128 (d))))); w--; } @@ -1934,21 +1900,23 @@ core_combine_out_ca_sse2 (uint32_t * pd, m = *pm++; d = *pd; - *pd++ = pack_1x64_32 ( - pix_multiply_1x64 ( - pix_multiply_1x64 ( - unpack_32_1x64 (s), unpack_32_1x64 (m)), - negate_1x64 (expand_alpha_1x64 (unpack_32_1x64 (d))))); + *pd++ = pack_1x128_32 ( + pix_multiply_1x128 ( + pix_multiply_1x128 ( + unpack_32_1x128 (s), unpack_32_1x128 (m)), + negate_1x128 (expand_alpha_1x128 (unpack_32_1x128 (d))))); w--; } } -static force_inline void -core_combine_out_reverse_ca_sse2 (uint32_t * pd, - const uint32_t *ps, - const uint32_t *pm, - int w) +static void +sse2_combine_out_reverse_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * pd, + const uint32_t * ps, + const uint32_t * pm, + int w) { uint32_t s, m, d; @@ -1963,12 +1931,12 @@ core_combine_out_reverse_ca_sse2 (uint32_t * pd, m = *pm++; d = *pd; - *pd++ = pack_1x64_32 ( - pix_multiply_1x64 ( - unpack_32_1x64 (d), - negate_1x64 (pix_multiply_1x64 ( - unpack_32_1x64 (m), - expand_alpha_1x64 (unpack_32_1x64 (s)))))); + *pd++ = pack_1x128_32 ( + pix_multiply_1x128 ( + unpack_32_1x128 (d), + negate_1x128 (pix_multiply_1x128 ( + unpack_32_1x128 (m), + expand_alpha_1x128 (unpack_32_1x128 (s)))))); w--; } @@ -2011,12 +1979,12 @@ core_combine_out_reverse_ca_sse2 (uint32_t * pd, m = *pm++; d = *pd; - *pd++ = pack_1x64_32 ( - pix_multiply_1x64 ( - unpack_32_1x64 (d), - negate_1x64 (pix_multiply_1x64 ( - unpack_32_1x64 (m), - expand_alpha_1x64 (unpack_32_1x64 (s)))))); + *pd++ = pack_1x128_32 ( + pix_multiply_1x128 ( + unpack_32_1x128 (d), + negate_1x128 (pix_multiply_1x128 ( + unpack_32_1x128 (m), + expand_alpha_1x128 (unpack_32_1x128 (s)))))); w--; } } @@ -2026,23 +1994,25 @@ core_combine_atop_ca_pixel_sse2 (uint32_t src, uint32_t mask, uint32_t dst) { - __m64 m = unpack_32_1x64 (mask); - __m64 s = unpack_32_1x64 (src); - __m64 d = unpack_32_1x64 (dst); - __m64 sa = expand_alpha_1x64 (s); - __m64 da = expand_alpha_1x64 (d); + __m128i m = unpack_32_1x128 (mask); + __m128i s = unpack_32_1x128 (src); + __m128i d = unpack_32_1x128 (dst); + __m128i sa = expand_alpha_1x128 (s); + __m128i da = expand_alpha_1x128 (d); - s = pix_multiply_1x64 (s, m); - m = negate_1x64 (pix_multiply_1x64 (m, sa)); + s = pix_multiply_1x128 (s, m); + m = negate_1x128 (pix_multiply_1x128 (m, sa)); - return pack_1x64_32 (pix_add_multiply_1x64 (&d, &m, &s, &da)); + return pack_1x128_32 (pix_add_multiply_1x128 (&d, &m, &s, &da)); } -static force_inline void -core_combine_atop_ca_sse2 (uint32_t * pd, - const uint32_t *ps, - const uint32_t *pm, - int w) +static void +sse2_combine_atop_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * pd, + const uint32_t * ps, + const uint32_t * pm, + int w) { uint32_t s, m, d; @@ -2116,24 +2086,26 @@ core_combine_reverse_atop_ca_pixel_sse2 (uint32_t src, uint32_t mask, uint32_t dst) { - __m64 m = unpack_32_1x64 (mask); - __m64 s = unpack_32_1x64 (src); - __m64 d = unpack_32_1x64 (dst); + __m128i m = unpack_32_1x128 (mask); + __m128i s = unpack_32_1x128 (src); + __m128i d = unpack_32_1x128 (dst); - __m64 da = negate_1x64 (expand_alpha_1x64 (d)); - __m64 sa = expand_alpha_1x64 (s); + __m128i da = negate_1x128 (expand_alpha_1x128 (d)); + __m128i sa = expand_alpha_1x128 (s); - s = pix_multiply_1x64 (s, m); - m = pix_multiply_1x64 (m, sa); + s = pix_multiply_1x128 (s, m); + m = pix_multiply_1x128 (m, sa); - return pack_1x64_32 (pix_add_multiply_1x64 (&d, &m, &s, &da)); + return pack_1x128_32 (pix_add_multiply_1x128 (&d, &m, &s, &da)); } -static force_inline void -core_combine_reverse_atop_ca_sse2 (uint32_t * pd, - const uint32_t *ps, - const uint32_t *pm, - int w) +static void +sse2_combine_atop_reverse_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * pd, + const uint32_t * ps, + const uint32_t * pm, + int w) { uint32_t s, m, d; @@ -2208,26 +2180,28 @@ core_combine_xor_ca_pixel_sse2 (uint32_t src, uint32_t mask, uint32_t dst) { - __m64 a = unpack_32_1x64 (mask); - __m64 s = unpack_32_1x64 (src); - __m64 d = unpack_32_1x64 (dst); + __m128i a = unpack_32_1x128 (mask); + __m128i s = unpack_32_1x128 (src); + __m128i d = unpack_32_1x128 (dst); - __m64 alpha_dst = negate_1x64 (pix_multiply_1x64 ( - a, expand_alpha_1x64 (s))); - __m64 dest = pix_multiply_1x64 (s, a); - __m64 alpha_src = negate_1x64 (expand_alpha_1x64 (d)); + __m128i alpha_dst = negate_1x128 (pix_multiply_1x128 ( + a, expand_alpha_1x128 (s))); + __m128i dest = pix_multiply_1x128 (s, a); + __m128i alpha_src = negate_1x128 (expand_alpha_1x128 (d)); - return pack_1x64_32 (pix_add_multiply_1x64 (&d, + return pack_1x128_32 (pix_add_multiply_1x128 (&d, &alpha_dst, &dest, &alpha_src)); } -static force_inline void -core_combine_xor_ca_sse2 (uint32_t * pd, - const uint32_t *ps, - const uint32_t *pm, - int w) +static void +sse2_combine_xor_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * pd, + const uint32_t * ps, + const uint32_t * pm, + int w) { uint32_t s, m, d; @@ -2299,11 +2273,13 @@ core_combine_xor_ca_sse2 (uint32_t * pd, } } -static force_inline void -core_combine_add_ca_sse2 (uint32_t * pd, - const uint32_t *ps, - const uint32_t *pm, - int w) +static void +sse2_combine_add_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * pd, + const uint32_t * ps, + const uint32_t * pm, + int w) { uint32_t s, m, d; @@ -2317,10 +2293,10 @@ core_combine_add_ca_sse2 (uint32_t * pd, m = *pm++; d = *pd; - *pd++ = pack_1x64_32 ( - _mm_adds_pu8 (pix_multiply_1x64 (unpack_32_1x64 (s), - unpack_32_1x64 (m)), - unpack_32_1x64 (d))); + *pd++ = pack_1x128_32 ( + _mm_adds_epu8 (pix_multiply_1x128 (unpack_32_1x128 (s), + unpack_32_1x128 (m)), + unpack_32_1x128 (d))); w--; } @@ -2355,36 +2331,20 @@ core_combine_add_ca_sse2 (uint32_t * pd, m = *pm++; d = *pd; - *pd++ = pack_1x64_32 ( - _mm_adds_pu8 (pix_multiply_1x64 (unpack_32_1x64 (s), - unpack_32_1x64 (m)), - unpack_32_1x64 (d))); + *pd++ = pack_1x128_32 ( + _mm_adds_epu8 (pix_multiply_1x128 (unpack_32_1x128 (s), + unpack_32_1x128 (m)), + unpack_32_1x128 (d))); w--; } } -/* --------------------------------------------------- - * fb_compose_setup_sSE2 - */ -static force_inline __m64 -create_mask_16_64 (uint16_t mask) -{ - return _mm_set1_pi16 (mask); -} - static force_inline __m128i create_mask_16_128 (uint16_t mask) { return _mm_set1_epi16 (mask); } -static force_inline __m64 -create_mask_2x32_64 (uint32_t mask0, - uint32_t mask1) -{ - return _mm_set_pi32 (mask0, mask1); -} - /* Work around a code generation bug in Sun Studio 12. */ #if defined(__SUNPRO_C) && (__SUNPRO_C >= 0x590) # define create_mask_2x32_128(mask0, mask1) \ @@ -2398,276 +2358,6 @@ create_mask_2x32_128 (uint32_t mask0, } #endif -/* SSE2 code patch for fbcompose.c */ - -static void -sse2_combine_over_u (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dst, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - core_combine_over_u_sse2 (dst, src, mask, width); - _mm_empty (); -} - -static void -sse2_combine_over_reverse_u (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dst, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - core_combine_over_reverse_u_sse2 (dst, src, mask, width); - _mm_empty (); -} - -static void -sse2_combine_in_u (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dst, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - core_combine_in_u_sse2 (dst, src, mask, width); - _mm_empty (); -} - -static void -sse2_combine_in_reverse_u (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dst, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - core_combine_reverse_in_u_sse2 (dst, src, mask, width); - _mm_empty (); -} - -static void -sse2_combine_out_u (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dst, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - core_combine_out_u_sse2 (dst, src, mask, width); - _mm_empty (); -} - -static void -sse2_combine_out_reverse_u (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dst, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - core_combine_reverse_out_u_sse2 (dst, src, mask, width); - _mm_empty (); -} - -static void -sse2_combine_atop_u (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dst, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - core_combine_atop_u_sse2 (dst, src, mask, width); - _mm_empty (); -} - -static void -sse2_combine_atop_reverse_u (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dst, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - core_combine_reverse_atop_u_sse2 (dst, src, mask, width); - _mm_empty (); -} - -static void -sse2_combine_xor_u (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dst, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - core_combine_xor_u_sse2 (dst, src, mask, width); - _mm_empty (); -} - -static void -sse2_combine_add_u (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dst, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - core_combine_add_u_sse2 (dst, src, mask, width); - _mm_empty (); -} - -static void -sse2_combine_saturate_u (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dst, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - core_combine_saturate_u_sse2 (dst, src, mask, width); - _mm_empty (); -} - -static void -sse2_combine_src_ca (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dst, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - core_combine_src_ca_sse2 (dst, src, mask, width); - _mm_empty (); -} - -static void -sse2_combine_over_ca (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dst, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - core_combine_over_ca_sse2 (dst, src, mask, width); - _mm_empty (); -} - -static void -sse2_combine_over_reverse_ca (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dst, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - core_combine_over_reverse_ca_sse2 (dst, src, mask, width); - _mm_empty (); -} - -static void -sse2_combine_in_ca (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dst, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - core_combine_in_ca_sse2 (dst, src, mask, width); - _mm_empty (); -} - -static void -sse2_combine_in_reverse_ca (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dst, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - core_combine_in_reverse_ca_sse2 (dst, src, mask, width); - _mm_empty (); -} - -static void -sse2_combine_out_ca (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dst, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - core_combine_out_ca_sse2 (dst, src, mask, width); - _mm_empty (); -} - -static void -sse2_combine_out_reverse_ca (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dst, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - core_combine_out_reverse_ca_sse2 (dst, src, mask, width); - _mm_empty (); -} - -static void -sse2_combine_atop_ca (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dst, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - core_combine_atop_ca_sse2 (dst, src, mask, width); - _mm_empty (); -} - -static void -sse2_combine_atop_reverse_ca (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dst, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - core_combine_reverse_atop_ca_sse2 (dst, src, mask, width); - _mm_empty (); -} - -static void -sse2_combine_xor_ca (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dst, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - core_combine_xor_ca_sse2 (dst, src, mask, width); - _mm_empty (); -} - -static void -sse2_combine_add_ca (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dst, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - core_combine_add_ca_sse2 (dst, src, mask, width); - _mm_empty (); -} - -/* ------------------------------------------------------------------- - * composite_over_n_8888 - */ - static void sse2_composite_over_n_8888 (pixman_implementation_t *imp, pixman_op_t op, @@ -2711,9 +2401,9 @@ sse2_composite_over_n_8888 (pixman_implementation_t *imp, while (w && (unsigned long)dst & 15) { d = *dst; - *dst++ = pack_1x64_32 (over_1x64 (_mm_movepi64_pi64 (xmm_src), - _mm_movepi64_pi64 (xmm_alpha), - unpack_32_1x64 (d))); + *dst++ = pack_1x128_32 (over_1x128 (xmm_src, + xmm_alpha, + unpack_32_1x128 (d))); w--; } @@ -2738,19 +2428,15 @@ sse2_composite_over_n_8888 (pixman_implementation_t *imp, while (w) { d = *dst; - *dst++ = pack_1x64_32 (over_1x64 (_mm_movepi64_pi64 (xmm_src), - _mm_movepi64_pi64 (xmm_alpha), - unpack_32_1x64 (d))); + *dst++ = pack_1x128_32 (over_1x128 (xmm_src, + xmm_alpha, + unpack_32_1x128 (d))); w--; } } - _mm_empty (); } -/* --------------------------------------------------------------------- - * composite_over_n_0565 - */ static void sse2_composite_over_n_0565 (pixman_implementation_t *imp, pixman_op_t op, @@ -2796,9 +2482,9 @@ sse2_composite_over_n_0565 (pixman_implementation_t *imp, d = *dst; *dst++ = pack_565_32_16 ( - pack_1x64_32 (over_1x64 (_mm_movepi64_pi64 (xmm_src), - _mm_movepi64_pi64 (xmm_alpha), - expand565_16_1x64 (d)))); + pack_1x128_32 (over_1x128 (xmm_src, + xmm_alpha, + expand565_16_1x128 (d)))); w--; } @@ -2829,18 +2515,13 @@ sse2_composite_over_n_0565 (pixman_implementation_t *imp, { d = *dst; *dst++ = pack_565_32_16 ( - pack_1x64_32 (over_1x64 (_mm_movepi64_pi64 (xmm_src), - _mm_movepi64_pi64 (xmm_alpha), - expand565_16_1x64 (d)))); + pack_1x128_32 (over_1x128 (xmm_src, xmm_alpha, + expand565_16_1x128 (d)))); } } - _mm_empty (); } -/* ------------------------------ - * composite_add_n_8888_8888_ca - */ static void sse2_composite_add_n_8888_8888_ca (pixman_implementation_t *imp, pixman_op_t op, @@ -2866,7 +2547,7 @@ sse2_composite_add_n_8888_8888_ca (pixman_implementation_t *imp, __m128i xmm_dst; __m128i xmm_mask, xmm_mask_lo, xmm_mask_hi; - __m64 mmx_src, mmx_alpha, mmx_mask, mmx_dest; + __m128i mmx_src, mmx_alpha, mmx_mask, mmx_dest; src = _pixman_image_get_solid (imp, src_image, dst_image->bits.format); srca = src >> 24; @@ -2882,8 +2563,8 @@ sse2_composite_add_n_8888_8888_ca (pixman_implementation_t *imp, xmm_src = _mm_unpacklo_epi8 ( create_mask_2x32_128 (src, src), _mm_setzero_si128 ()); xmm_alpha = expand_alpha_1x128 (xmm_src); - mmx_src = _mm_movepi64_pi64 (xmm_src); - mmx_alpha = _mm_movepi64_pi64 (xmm_alpha); + mmx_src = xmm_src; + mmx_alpha = xmm_alpha; while (height--) { @@ -2902,11 +2583,12 @@ sse2_composite_add_n_8888_8888_ca (pixman_implementation_t *imp, { d = *pd; - mmx_mask = unpack_32_1x64 (m); - mmx_dest = unpack_32_1x64 (d); + mmx_mask = unpack_32_1x128 (m); + mmx_dest = unpack_32_1x128 (d); - *pd = pack_1x64_32 ( - _mm_adds_pu8 (pix_multiply_1x64 (mmx_mask, mmx_src), mmx_dest)); + *pd = pack_1x128_32 ( + _mm_adds_epu8 (pix_multiply_1x128 (mmx_mask, mmx_src), + mmx_dest)); } pd++; @@ -2950,11 +2632,12 @@ sse2_composite_add_n_8888_8888_ca (pixman_implementation_t *imp, { d = *pd; - mmx_mask = unpack_32_1x64 (m); - mmx_dest = unpack_32_1x64 (d); + mmx_mask = unpack_32_1x128 (m); + mmx_dest = unpack_32_1x128 (d); - *pd = pack_1x64_32 ( - _mm_adds_pu8 (pix_multiply_1x64 (mmx_mask, mmx_src), mmx_dest)); + *pd = pack_1x128_32 ( + _mm_adds_epu8 (pix_multiply_1x128 (mmx_mask, mmx_src), + mmx_dest)); } pd++; @@ -2962,13 +2645,8 @@ sse2_composite_add_n_8888_8888_ca (pixman_implementation_t *imp, } } - _mm_empty (); } -/* --------------------------------------------------------------------------- - * composite_over_n_8888_8888_ca - */ - static void sse2_composite_over_n_8888_8888_ca (pixman_implementation_t *imp, pixman_op_t op, @@ -2994,7 +2672,7 @@ sse2_composite_over_n_8888_8888_ca (pixman_implementation_t *imp, __m128i xmm_dst, xmm_dst_lo, xmm_dst_hi; __m128i xmm_mask, xmm_mask_lo, xmm_mask_hi; - __m64 mmx_src, mmx_alpha, mmx_mask, mmx_dest; + __m128i mmx_src, mmx_alpha, mmx_mask, mmx_dest; src = _pixman_image_get_solid (imp, src_image, dst_image->bits.format); @@ -3009,8 +2687,8 @@ sse2_composite_over_n_8888_8888_ca (pixman_implementation_t *imp, xmm_src = _mm_unpacklo_epi8 ( create_mask_2x32_128 (src, src), _mm_setzero_si128 ()); xmm_alpha = expand_alpha_1x128 (xmm_src); - mmx_src = _mm_movepi64_pi64 (xmm_src); - mmx_alpha = _mm_movepi64_pi64 (xmm_alpha); + mmx_src = xmm_src; + mmx_alpha = xmm_alpha; while (height--) { @@ -3028,10 +2706,10 @@ sse2_composite_over_n_8888_8888_ca (pixman_implementation_t *imp, if (m) { d = *pd; - mmx_mask = unpack_32_1x64 (m); - mmx_dest = unpack_32_1x64 (d); + mmx_mask = unpack_32_1x128 (m); + mmx_dest = unpack_32_1x128 (d); - *pd = pack_1x64_32 (in_over_1x64 (&mmx_src, + *pd = pack_1x128_32 (in_over_1x128 (&mmx_src, &mmx_alpha, &mmx_mask, &mmx_dest)); @@ -3078,11 +2756,11 @@ sse2_composite_over_n_8888_8888_ca (pixman_implementation_t *imp, if (m) { d = *pd; - mmx_mask = unpack_32_1x64 (m); - mmx_dest = unpack_32_1x64 (d); + mmx_mask = unpack_32_1x128 (m); + mmx_dest = unpack_32_1x128 (d); - *pd = pack_1x64_32 ( - in_over_1x64 (&mmx_src, &mmx_alpha, &mmx_mask, &mmx_dest)); + *pd = pack_1x128_32 ( + in_over_1x128 (&mmx_src, &mmx_alpha, &mmx_mask, &mmx_dest)); } pd++; @@ -3090,13 +2768,8 @@ sse2_composite_over_n_8888_8888_ca (pixman_implementation_t *imp, } } - _mm_empty (); } -/*--------------------------------------------------------------------- - * composite_over_8888_n_8888 - */ - static void sse2_composite_over_8888_n_8888 (pixman_implementation_t *imp, pixman_op_t op, @@ -3148,13 +2821,13 @@ sse2_composite_over_8888_n_8888 (pixman_implementation_t *imp, { uint32_t d = *dst; - __m64 ms = unpack_32_1x64 (s); - __m64 alpha = expand_alpha_1x64 (ms); - __m64 dest = _mm_movepi64_pi64 (xmm_mask); - __m64 alpha_dst = unpack_32_1x64 (d); + __m128i ms = unpack_32_1x128 (s); + __m128i alpha = expand_alpha_1x128 (ms); + __m128i dest = xmm_mask; + __m128i alpha_dst = unpack_32_1x128 (d); - *dst = pack_1x64_32 ( - in_over_1x64 (&ms, &alpha, &dest, &alpha_dst)); + *dst = pack_1x128_32 ( + in_over_1x128 (&ms, &alpha, &dest, &alpha_dst)); } dst++; w--; @@ -3195,13 +2868,13 @@ sse2_composite_over_8888_n_8888 (pixman_implementation_t *imp, { uint32_t d = *dst; - __m64 ms = unpack_32_1x64 (s); - __m64 alpha = expand_alpha_1x64 (ms); - __m64 mask = _mm_movepi64_pi64 (xmm_mask); - __m64 dest = unpack_32_1x64 (d); + __m128i ms = unpack_32_1x128 (s); + __m128i alpha = expand_alpha_1x128 (ms); + __m128i mask = xmm_mask; + __m128i dest = unpack_32_1x128 (d); - *dst = pack_1x64_32 ( - in_over_1x64 (&ms, &alpha, &mask, &dest)); + *dst = pack_1x128_32 ( + in_over_1x128 (&ms, &alpha, &mask, &dest)); } dst++; @@ -3209,13 +2882,8 @@ sse2_composite_over_8888_n_8888 (pixman_implementation_t *imp, } } - _mm_empty (); } -/*--------------------------------------------------------------------- - * composite_over_8888_n_8888 - */ - static void sse2_composite_src_x888_8888 (pixman_implementation_t *imp, pixman_op_t op, @@ -3282,12 +2950,8 @@ sse2_composite_src_x888_8888 (pixman_implementation_t *imp, } } - _mm_empty (); } -/* --------------------------------------------------------------------- - * composite_over_x888_n_8888 - */ static void sse2_composite_over_x888_n_8888 (pixman_implementation_t *imp, pixman_op_t op, @@ -3336,13 +3000,13 @@ sse2_composite_over_x888_n_8888 (pixman_implementation_t *imp, uint32_t s = (*src++) | 0xff000000; uint32_t d = *dst; - __m64 src = unpack_32_1x64 (s); - __m64 alpha = _mm_movepi64_pi64 (xmm_alpha); - __m64 mask = _mm_movepi64_pi64 (xmm_mask); - __m64 dest = unpack_32_1x64 (d); + __m128i src = unpack_32_1x128 (s); + __m128i alpha = xmm_alpha; + __m128i mask = xmm_mask; + __m128i dest = unpack_32_1x128 (d); - *dst++ = pack_1x64_32 ( - in_over_1x64 (&src, &alpha, &mask, &dest)); + *dst++ = pack_1x128_32 ( + in_over_1x128 (&src, &alpha, &mask, &dest)); w--; } @@ -3375,24 +3039,20 @@ sse2_composite_over_x888_n_8888 (pixman_implementation_t *imp, uint32_t s = (*src++) | 0xff000000; uint32_t d = *dst; - __m64 src = unpack_32_1x64 (s); - __m64 alpha = _mm_movepi64_pi64 (xmm_alpha); - __m64 mask = _mm_movepi64_pi64 (xmm_mask); - __m64 dest = unpack_32_1x64 (d); + __m128i src = unpack_32_1x128 (s); + __m128i alpha = xmm_alpha; + __m128i mask = xmm_mask; + __m128i dest = unpack_32_1x128 (d); - *dst++ = pack_1x64_32 ( - in_over_1x64 (&src, &alpha, &mask, &dest)); + *dst++ = pack_1x128_32 ( + in_over_1x128 (&src, &alpha, &mask, &dest)); w--; } } - _mm_empty (); } -/* -------------------------------------------------------------------- - * composite_over_8888_8888 - */ static void sse2_composite_over_8888_8888 (pixman_implementation_t *imp, pixman_op_t op, @@ -3422,27 +3082,23 @@ sse2_composite_over_8888_8888 (pixman_implementation_t *imp, while (height--) { - core_combine_over_u_sse2 (dst, src, NULL, width); + sse2_combine_over_u (imp, op, dst, src, NULL, width); dst += dst_stride; src += src_stride; } - _mm_empty (); } -/* ------------------------------------------------------------------ - * composite_over_8888_0565 - */ static force_inline uint16_t composite_over_8888_0565pixel (uint32_t src, uint16_t dst) { - __m64 ms; + __m128i ms; - ms = unpack_32_1x64 (src); + ms = unpack_32_1x128 (src); return pack_565_32_16 ( - pack_1x64_32 ( - over_1x64 ( - ms, expand_alpha_1x64 (ms), expand565_16_1x64 (dst)))); + pack_1x128_32 ( + over_1x128 ( + ms, expand_alpha_1x128 (ms), expand565_16_1x128 (dst)))); } static void @@ -3474,15 +3130,6 @@ sse2_composite_over_8888_0565 (pixman_implementation_t *imp, PIXMAN_IMAGE_GET_LINE ( src_image, src_x, src_y, uint32_t, src_stride, src_line, 1); -#if 0 - /* FIXME - * - * I copy the code from MMX one and keep the fixme. - * If it's a problem there, probably is a problem here. - */ - assert (src_image->drawable == mask_image->drawable); -#endif - while (height--) { dst = dst_line; @@ -3555,13 +3202,8 @@ sse2_composite_over_8888_0565 (pixman_implementation_t *imp, } } - _mm_empty (); } -/* ----------------------------------------------------------------- - * composite_over_n_8_8888 - */ - static void sse2_composite_over_n_8_8888 (pixman_implementation_t *imp, pixman_op_t op, @@ -3588,7 +3230,7 @@ sse2_composite_over_n_8_8888 (pixman_implementation_t *imp, __m128i xmm_dst, xmm_dst_lo, xmm_dst_hi; __m128i xmm_mask, xmm_mask_lo, xmm_mask_hi; - __m64 mmx_src, mmx_alpha, mmx_mask, mmx_dest; + __m128i mmx_src, mmx_alpha, mmx_mask, mmx_dest; src = _pixman_image_get_solid (imp, src_image, dst_image->bits.format); @@ -3604,8 +3246,8 @@ sse2_composite_over_n_8_8888 (pixman_implementation_t *imp, xmm_def = create_mask_2x32_128 (src, src); xmm_src = expand_pixel_32_1x128 (src); xmm_alpha = expand_alpha_1x128 (xmm_src); - mmx_src = _mm_movepi64_pi64 (xmm_src); - mmx_alpha = _mm_movepi64_pi64 (xmm_alpha); + mmx_src = xmm_src; + mmx_alpha = xmm_alpha; while (height--) { @@ -3622,10 +3264,10 @@ sse2_composite_over_n_8_8888 (pixman_implementation_t *imp, if (m) { d = *dst; - mmx_mask = expand_pixel_8_1x64 (m); - mmx_dest = unpack_32_1x64 (d); + mmx_mask = expand_pixel_8_1x128 (m); + mmx_dest = unpack_32_1x128 (d); - *dst = pack_1x64_32 (in_over_1x64 (&mmx_src, + *dst = pack_1x128_32 (in_over_1x128 (&mmx_src, &mmx_alpha, &mmx_mask, &mmx_dest)); @@ -3677,10 +3319,10 @@ sse2_composite_over_n_8_8888 (pixman_implementation_t *imp, if (m) { d = *dst; - mmx_mask = expand_pixel_8_1x64 (m); - mmx_dest = unpack_32_1x64 (d); + mmx_mask = expand_pixel_8_1x128 (m); + mmx_dest = unpack_32_1x128 (d); - *dst = pack_1x64_32 (in_over_1x64 (&mmx_src, + *dst = pack_1x128_32 (in_over_1x128 (&mmx_src, &mmx_alpha, &mmx_mask, &mmx_dest)); @@ -3691,14 +3333,9 @@ sse2_composite_over_n_8_8888 (pixman_implementation_t *imp, } } - _mm_empty (); } -/* ---------------------------------------------------------------- - * composite_over_n_8_8888 - */ - -pixman_bool_t +static pixman_bool_t pixman_fill_sse2 (uint32_t *bits, int stride, int bpp, @@ -3845,7 +3482,6 @@ pixman_fill_sse2 (uint32_t *bits, } } - _mm_empty (); return TRUE; } @@ -3907,9 +3543,8 @@ sse2_composite_src_n_8_8888 (pixman_implementation_t *imp, if (m) { - *dst = pack_1x64_32 ( - pix_multiply_1x64 ( - _mm_movepi64_pi64 (xmm_src), expand_pixel_8_1x64 (m))); + *dst = pack_1x128_32 ( + pix_multiply_1x128 (xmm_src, expand_pixel_8_1x128 (m))); } else { @@ -3962,9 +3597,9 @@ sse2_composite_src_n_8_8888 (pixman_implementation_t *imp, if (m) { - *dst = pack_1x64_32 ( - pix_multiply_1x64 ( - _mm_movepi64_pi64 (xmm_src), expand_pixel_8_1x64 (m))); + *dst = pack_1x128_32 ( + pix_multiply_1x128 ( + xmm_src, expand_pixel_8_1x128 (m))); } else { @@ -3976,13 +3611,8 @@ sse2_composite_src_n_8_8888 (pixman_implementation_t *imp, } } - _mm_empty (); } -/*----------------------------------------------------------------------- - * composite_over_n_8_0565 - */ - static void sse2_composite_over_n_8_0565 (pixman_implementation_t *imp, pixman_op_t op, @@ -4004,7 +3634,7 @@ sse2_composite_over_n_8_0565 (pixman_implementation_t *imp, int dst_stride, mask_stride; int32_t w; uint32_t m; - __m64 mmx_src, mmx_alpha, mmx_mask, mmx_dest; + __m128i mmx_src, mmx_alpha, mmx_mask, mmx_dest; __m128i xmm_src, xmm_alpha; __m128i xmm_mask, xmm_mask_lo, xmm_mask_hi; @@ -4023,8 +3653,8 @@ sse2_composite_over_n_8_0565 (pixman_implementation_t *imp, xmm_src = expand_pixel_32_1x128 (src); xmm_alpha = expand_alpha_1x128 (xmm_src); - mmx_src = _mm_movepi64_pi64 (xmm_src); - mmx_alpha = _mm_movepi64_pi64 (xmm_alpha); + mmx_src = xmm_src; + mmx_alpha = xmm_alpha; while (height--) { @@ -4041,12 +3671,12 @@ sse2_composite_over_n_8_0565 (pixman_implementation_t *imp, if (m) { d = *dst; - mmx_mask = expand_alpha_rev_1x64 (unpack_32_1x64 (m)); - mmx_dest = expand565_16_1x64 (d); + mmx_mask = expand_alpha_rev_1x128 (unpack_32_1x128 (m)); + mmx_dest = expand565_16_1x128 (d); *dst = pack_565_32_16 ( - pack_1x64_32 ( - in_over_1x64 ( + pack_1x128_32 ( + in_over_1x128 ( &mmx_src, &mmx_alpha, &mmx_mask, &mmx_dest))); } @@ -4114,12 +3744,12 @@ sse2_composite_over_n_8_0565 (pixman_implementation_t *imp, if (m) { d = *dst; - mmx_mask = expand_alpha_rev_1x64 (unpack_32_1x64 (m)); - mmx_dest = expand565_16_1x64 (d); + mmx_mask = expand_alpha_rev_1x128 (unpack_32_1x128 (m)); + mmx_dest = expand565_16_1x128 (d); *dst = pack_565_32_16 ( - pack_1x64_32 ( - in_over_1x64 ( + pack_1x128_32 ( + in_over_1x128 ( &mmx_src, &mmx_alpha, &mmx_mask, &mmx_dest))); } @@ -4128,13 +3758,8 @@ sse2_composite_over_n_8_0565 (pixman_implementation_t *imp, } } - _mm_empty (); } -/* ----------------------------------------------------------------------- - * composite_over_pixbuf_0565 - */ - static void sse2_composite_over_pixbuf_0565 (pixman_implementation_t *imp, pixman_op_t op, @@ -4156,7 +3781,7 @@ sse2_composite_over_pixbuf_0565 (pixman_implementation_t *imp, int32_t w; uint32_t opaque, zero; - __m64 ms; + __m128i ms; __m128i xmm_src, xmm_src_lo, xmm_src_hi; __m128i xmm_dst, xmm_dst0, xmm_dst1, xmm_dst2, xmm_dst3; @@ -4165,15 +3790,6 @@ sse2_composite_over_pixbuf_0565 (pixman_implementation_t *imp, PIXMAN_IMAGE_GET_LINE ( src_image, src_x, src_y, uint32_t, src_stride, src_line, 1); -#if 0 - /* FIXME - * - * I copy the code from MMX one and keep the fixme. - * If it's a problem there, probably is a problem here. - */ - assert (src_image->drawable == mask_image->drawable); -#endif - while (height--) { dst = dst_line; @@ -4187,11 +3803,11 @@ sse2_composite_over_pixbuf_0565 (pixman_implementation_t *imp, s = *src++; d = *dst; - ms = unpack_32_1x64 (s); + ms = unpack_32_1x128 (s); *dst++ = pack_565_32_16 ( - pack_1x64_32 ( - over_rev_non_pre_1x64 (ms, expand565_16_1x64 (d)))); + pack_1x128_32 ( + over_rev_non_pre_1x128 (ms, expand565_16_1x128 (d)))); w--; } @@ -4253,22 +3869,17 @@ sse2_composite_over_pixbuf_0565 (pixman_implementation_t *imp, s = *src++; d = *dst; - ms = unpack_32_1x64 (s); + ms = unpack_32_1x128 (s); *dst++ = pack_565_32_16 ( - pack_1x64_32 ( - over_rev_non_pre_1x64 (ms, expand565_16_1x64 (d)))); + pack_1x128_32 ( + over_rev_non_pre_1x128 (ms, expand565_16_1x128 (d)))); w--; } } - _mm_empty (); } -/* ------------------------------------------------------------------------- - * composite_over_pixbuf_8888 - */ - static void sse2_composite_over_pixbuf_8888 (pixman_implementation_t *imp, pixman_op_t op, @@ -4298,15 +3909,6 @@ sse2_composite_over_pixbuf_8888 (pixman_implementation_t *imp, PIXMAN_IMAGE_GET_LINE ( src_image, src_x, src_y, uint32_t, src_stride, src_line, 1); -#if 0 - /* FIXME - * - * I copy the code from MMX one and keep the fixme. - * If it's a problem there, probably is a problem here. - */ - assert (src_image->drawable == mask_image->drawable); -#endif - while (height--) { dst = dst_line; @@ -4320,9 +3922,9 @@ sse2_composite_over_pixbuf_8888 (pixman_implementation_t *imp, s = *src++; d = *dst; - *dst++ = pack_1x64_32 ( - over_rev_non_pre_1x64 ( - unpack_32_1x64 (s), unpack_32_1x64 (d))); + *dst++ = pack_1x128_32 ( + over_rev_non_pre_1x128 ( + unpack_32_1x128 (s), unpack_32_1x128 (d))); w--; } @@ -4367,21 +3969,16 @@ sse2_composite_over_pixbuf_8888 (pixman_implementation_t *imp, s = *src++; d = *dst; - *dst++ = pack_1x64_32 ( - over_rev_non_pre_1x64 ( - unpack_32_1x64 (s), unpack_32_1x64 (d))); + *dst++ = pack_1x128_32 ( + over_rev_non_pre_1x128 ( + unpack_32_1x128 (s), unpack_32_1x128 (d))); w--; } } - _mm_empty (); } -/* ------------------------------------------------------------------------------------------------- - * composite_over_n_8888_0565_ca - */ - static void sse2_composite_over_n_8888_0565_ca (pixman_implementation_t *imp, pixman_op_t op, @@ -4408,7 +4005,7 @@ sse2_composite_over_n_8888_0565_ca (pixman_implementation_t *imp, __m128i xmm_mask, xmm_mask_lo, xmm_mask_hi; __m128i xmm_dst, xmm_dst0, xmm_dst1, xmm_dst2, xmm_dst3; - __m64 mmx_src, mmx_alpha, mmx_mask, mmx_dest; + __m128i mmx_src, mmx_alpha, mmx_mask, mmx_dest; src = _pixman_image_get_solid (imp, src_image, dst_image->bits.format); @@ -4422,8 +4019,8 @@ sse2_composite_over_n_8888_0565_ca (pixman_implementation_t *imp, xmm_src = expand_pixel_32_1x128 (src); xmm_alpha = expand_alpha_1x128 (xmm_src); - mmx_src = _mm_movepi64_pi64 (xmm_src); - mmx_alpha = _mm_movepi64_pi64 (xmm_alpha); + mmx_src = xmm_src; + mmx_alpha = xmm_alpha; while (height--) { @@ -4440,12 +4037,12 @@ sse2_composite_over_n_8888_0565_ca (pixman_implementation_t *imp, if (m) { d = *dst; - mmx_mask = unpack_32_1x64 (m); - mmx_dest = expand565_16_1x64 (d); + mmx_mask = unpack_32_1x128 (m); + mmx_dest = expand565_16_1x128 (d); *dst = pack_565_32_16 ( - pack_1x64_32 ( - in_over_1x64 ( + pack_1x128_32 ( + in_over_1x128 ( &mmx_src, &mmx_alpha, &mmx_mask, &mmx_dest))); } @@ -4509,12 +4106,12 @@ sse2_composite_over_n_8888_0565_ca (pixman_implementation_t *imp, if (m) { d = *dst; - mmx_mask = unpack_32_1x64 (m); - mmx_dest = expand565_16_1x64 (d); + mmx_mask = unpack_32_1x128 (m); + mmx_dest = expand565_16_1x128 (d); *dst = pack_565_32_16 ( - pack_1x64_32 ( - in_over_1x64 ( + pack_1x128_32 ( + in_over_1x128 ( &mmx_src, &mmx_alpha, &mmx_mask, &mmx_dest))); } @@ -4524,13 +4121,8 @@ sse2_composite_over_n_8888_0565_ca (pixman_implementation_t *imp, } } - _mm_empty (); } -/* ----------------------------------------------------------------------- - * composite_in_n_8_8 - */ - static void sse2_composite_in_n_8_8 (pixman_implementation_t *imp, pixman_op_t op, @@ -4582,11 +4174,11 @@ sse2_composite_in_n_8_8 (pixman_implementation_t *imp, m = (uint32_t) *mask++; d = (uint32_t) *dst; - *dst++ = (uint8_t) pack_1x64_32 ( - pix_multiply_1x64 ( - pix_multiply_1x64 (_mm_movepi64_pi64 (xmm_alpha), - unpack_32_1x64 (m)), - unpack_32_1x64 (d))); + *dst++ = (uint8_t) pack_1x128_32 ( + pix_multiply_1x128 ( + pix_multiply_1x128 (xmm_alpha, + unpack_32_1x128 (m)), + unpack_32_1x128 (d))); w--; } @@ -4619,22 +4211,17 @@ sse2_composite_in_n_8_8 (pixman_implementation_t *imp, m = (uint32_t) *mask++; d = (uint32_t) *dst; - *dst++ = (uint8_t) pack_1x64_32 ( - pix_multiply_1x64 ( - pix_multiply_1x64 ( - _mm_movepi64_pi64 (xmm_alpha), unpack_32_1x64 (m)), - unpack_32_1x64 (d))); + *dst++ = (uint8_t) pack_1x128_32 ( + pix_multiply_1x128 ( + pix_multiply_1x128 ( + xmm_alpha, unpack_32_1x128 (m)), + unpack_32_1x128 (d))); w--; } } - _mm_empty (); } -/* ----------------------------------------------------------------------- - * composite_in_n_8 - */ - static void sse2_composite_in_n_8 (pixman_implementation_t *imp, pixman_op_t op, @@ -4689,10 +4276,10 @@ sse2_composite_in_n_8 (pixman_implementation_t *imp, { d = (uint32_t) *dst; - *dst++ = (uint8_t) pack_1x64_32 ( - pix_multiply_1x64 ( - _mm_movepi64_pi64 (xmm_alpha), - unpack_32_1x64 (d))); + *dst++ = (uint8_t) pack_1x128_32 ( + pix_multiply_1x128 ( + xmm_alpha, + unpack_32_1x128 (d))); w--; } @@ -4717,21 +4304,16 @@ sse2_composite_in_n_8 (pixman_implementation_t *imp, { d = (uint32_t) *dst; - *dst++ = (uint8_t) pack_1x64_32 ( - pix_multiply_1x64 ( - _mm_movepi64_pi64 (xmm_alpha), - unpack_32_1x64 (d))); + *dst++ = (uint8_t) pack_1x128_32 ( + pix_multiply_1x128 ( + xmm_alpha, + unpack_32_1x128 (d))); w--; } } - _mm_empty (); } -/* --------------------------------------------------------------------------- - * composite_in_8_8 - */ - static void sse2_composite_in_8_8 (pixman_implementation_t *imp, pixman_op_t op, @@ -4774,9 +4356,9 @@ sse2_composite_in_8_8 (pixman_implementation_t *imp, s = (uint32_t) *src++; d = (uint32_t) *dst; - *dst++ = (uint8_t) pack_1x64_32 ( - pix_multiply_1x64 ( - unpack_32_1x64 (s), unpack_32_1x64 (d))); + *dst++ = (uint8_t) pack_1x128_32 ( + pix_multiply_1x128 ( + unpack_32_1x128 (s), unpack_32_1x128 (d))); w--; } @@ -4805,19 +4387,14 @@ sse2_composite_in_8_8 (pixman_implementation_t *imp, s = (uint32_t) *src++; d = (uint32_t) *dst; - *dst++ = (uint8_t) pack_1x64_32 ( - pix_multiply_1x64 (unpack_32_1x64 (s), unpack_32_1x64 (d))); + *dst++ = (uint8_t) pack_1x128_32 ( + pix_multiply_1x128 (unpack_32_1x128 (s), unpack_32_1x128 (d))); w--; } } - _mm_empty (); } -/* ------------------------------------------------------------------------- - * composite_add_n_8_8 - */ - static void sse2_composite_add_n_8_8 (pixman_implementation_t *imp, pixman_op_t op, @@ -4869,11 +4446,11 @@ sse2_composite_add_n_8_8 (pixman_implementation_t *imp, m = (uint32_t) *mask++; d = (uint32_t) *dst; - *dst++ = (uint8_t) pack_1x64_32 ( - _mm_adds_pu16 ( - pix_multiply_1x64 ( - _mm_movepi64_pi64 (xmm_alpha), unpack_32_1x64 (m)), - unpack_32_1x64 (d))); + *dst++ = (uint8_t) pack_1x128_32 ( + _mm_adds_epu16 ( + pix_multiply_1x128 ( + xmm_alpha, unpack_32_1x128 (m)), + unpack_32_1x128 (d))); w--; } @@ -4905,23 +4482,18 @@ sse2_composite_add_n_8_8 (pixman_implementation_t *imp, m = (uint32_t) *mask++; d = (uint32_t) *dst; - *dst++ = (uint8_t) pack_1x64_32 ( - _mm_adds_pu16 ( - pix_multiply_1x64 ( - _mm_movepi64_pi64 (xmm_alpha), unpack_32_1x64 (m)), - unpack_32_1x64 (d))); + *dst++ = (uint8_t) pack_1x128_32 ( + _mm_adds_epu16 ( + pix_multiply_1x128 ( + xmm_alpha, unpack_32_1x128 (m)), + unpack_32_1x128 (d))); w--; } } - _mm_empty (); } -/* ------------------------------------------------------------------------- - * composite_add_n_8_8 - */ - static void sse2_composite_add_n_8 (pixman_implementation_t *imp, pixman_op_t op, @@ -4973,10 +4545,10 @@ sse2_composite_add_n_8 (pixman_implementation_t *imp, while (w && ((unsigned long)dst & 15)) { - *dst = (uint8_t)_mm_cvtsi64_si32 ( - _mm_adds_pu8 ( - _mm_movepi64_pi64 (xmm_src), - _mm_cvtsi32_si64 (*dst))); + *dst = (uint8_t)_mm_cvtsi128_si32 ( + _mm_adds_epu8 ( + xmm_src, + _mm_cvtsi32_si128 (*dst))); w--; dst++; @@ -4993,23 +4565,18 @@ sse2_composite_add_n_8 (pixman_implementation_t *imp, while (w) { - *dst = (uint8_t)_mm_cvtsi64_si32 ( - _mm_adds_pu8 ( - _mm_movepi64_pi64 (xmm_src), - _mm_cvtsi32_si64 (*dst))); + *dst = (uint8_t)_mm_cvtsi128_si32 ( + _mm_adds_epu8 ( + xmm_src, + _mm_cvtsi32_si128 (*dst))); w--; dst++; } } - _mm_empty (); } -/* ---------------------------------------------------------------------- - * composite_add_8_8 - */ - static void sse2_composite_add_8_8 (pixman_implementation_t *imp, pixman_op_t op, @@ -5053,7 +4620,8 @@ sse2_composite_add_8_8 (pixman_implementation_t *imp, w--; } - core_combine_add_u_sse2 ((uint32_t*)dst, (uint32_t*)src, NULL, w >> 2); + sse2_combine_add_u (imp, op, + (uint32_t*)dst, (uint32_t*)src, NULL, w >> 2); /* Small tail */ dst += w & 0xfffc; @@ -5069,12 +4637,8 @@ sse2_composite_add_8_8 (pixman_implementation_t *imp, } } - _mm_empty (); } -/* --------------------------------------------------------------------- - * composite_add_8888_8888 - */ static void sse2_composite_add_8888_8888 (pixman_implementation_t *imp, pixman_op_t op, @@ -5106,16 +4670,11 @@ sse2_composite_add_8888_8888 (pixman_implementation_t *imp, src = src_line; src_line += src_stride; - core_combine_add_u_sse2 (dst, src, NULL, width); + sse2_combine_add_u (imp, op, dst, src, NULL, width); } - _mm_empty (); } -/* ------------------------------------------------------------------------------------------------- - * sse2_composite_copy_area - */ - static pixman_bool_t pixman_blt_sse2 (uint32_t *src_bits, uint32_t *dst_bits, @@ -5234,7 +4793,6 @@ pixman_blt_sse2 (uint32_t *src_bits, } } - _mm_empty (); return TRUE; } @@ -5284,7 +4842,7 @@ sse2_composite_over_x888_8_8888 (pixman_implementation_t *imp, uint32_t m; int src_stride, mask_stride, dst_stride; int32_t w; - __m64 ms; + __m128i ms; __m128i xmm_src, xmm_src_lo, xmm_src_hi; __m128i xmm_dst, xmm_dst_lo, xmm_dst_hi; @@ -5313,24 +4871,25 @@ sse2_composite_over_x888_8_8888 (pixman_implementation_t *imp, s = 0xff000000 | *src++; m = (uint32_t) *mask++; d = *dst; - ms = unpack_32_1x64 (s); + ms = unpack_32_1x128 (s); if (m != 0xff) { - __m64 ma = expand_alpha_rev_1x64 (unpack_32_1x64 (m)); - __m64 md = unpack_32_1x64 (d); + __m128i ma = expand_alpha_rev_1x128 (unpack_32_1x128 (m)); + __m128i md = unpack_32_1x128 (d); - ms = in_over_1x64 (&ms, &mask_x00ff, &ma, &md); + ms = in_over_1x128 (&ms, &mask_00ff, &ma, &md); } - *dst++ = pack_1x64_32 (ms); + *dst++ = pack_1x128_32 (ms); w--; } while (w >= 4) { m = *(uint32_t*) mask; - xmm_src = _mm_or_si128 (load_128_unaligned ((__m128i*)src), mask_ff000000); + xmm_src = _mm_or_si128 ( + load_128_unaligned ((__m128i*)src), mask_ff000000); if (m == 0xffffffff) { @@ -5346,9 +4905,12 @@ sse2_composite_over_x888_8_8888 (pixman_implementation_t *imp, unpack_128_2x128 (xmm_mask, &xmm_mask_lo, &xmm_mask_hi); unpack_128_2x128 (xmm_dst, &xmm_dst_lo, &xmm_dst_hi); - expand_alpha_rev_2x128 (xmm_mask_lo, xmm_mask_hi, &xmm_mask_lo, &xmm_mask_hi); + expand_alpha_rev_2x128 ( + xmm_mask_lo, xmm_mask_hi, &xmm_mask_lo, &xmm_mask_hi); - in_over_2x128 (&xmm_src_lo, &xmm_src_hi, &mask_00ff, &mask_00ff, &xmm_mask_lo, &xmm_mask_hi, &xmm_dst_lo, &xmm_dst_hi); + in_over_2x128 (&xmm_src_lo, &xmm_src_hi, + &mask_00ff, &mask_00ff, &xmm_mask_lo, &xmm_mask_hi, + &xmm_dst_lo, &xmm_dst_hi); save_128_aligned ((__m128i*)dst, pack_2x128_128 (xmm_dst_lo, xmm_dst_hi)); } @@ -5373,15 +4935,15 @@ sse2_composite_over_x888_8_8888 (pixman_implementation_t *imp, } else { - __m64 ma, md, ms; + __m128i ma, md, ms; d = *dst; - ma = expand_alpha_rev_1x64 (unpack_32_1x64 (m)); - md = unpack_32_1x64 (d); - ms = unpack_32_1x64 (s); + ma = expand_alpha_rev_1x128 (unpack_32_1x128 (m)); + md = unpack_32_1x128 (d); + ms = unpack_32_1x128 (s); - *dst = pack_1x64_32 (in_over_1x64 (&ms, &mask_x00ff, &ma, &md)); + *dst = pack_1x128_32 (in_over_1x128 (&ms, &mask_00ff, &ma, &md)); } } @@ -5392,7 +4954,6 @@ sse2_composite_over_x888_8_8888 (pixman_implementation_t *imp, } } - _mm_empty (); } static void @@ -5457,15 +5018,15 @@ sse2_composite_over_8888_8_8888 (pixman_implementation_t *imp, } else { - __m64 ms, md, ma, msa; + __m128i ms, md, ma, msa; - ma = expand_alpha_rev_1x64 (load_32_1x64 (m)); - ms = unpack_32_1x64 (s); - md = unpack_32_1x64 (d); + ma = expand_alpha_rev_1x128 (load_32_1x128 (m)); + ms = unpack_32_1x128 (s); + md = unpack_32_1x128 (d); - msa = expand_alpha_rev_1x64 (load_32_1x64 (sa)); + msa = expand_alpha_rev_1x128 (load_32_1x128 (sa)); - *dst = pack_1x64_32 (in_over_1x64 (&ms, &msa, &ma, &md)); + *dst = pack_1x128_32 (in_over_1x128 (&ms, &msa, &ma, &md)); } } @@ -5529,15 +5090,15 @@ sse2_composite_over_8888_8_8888 (pixman_implementation_t *imp, } else { - __m64 ms, md, ma, msa; + __m128i ms, md, ma, msa; - ma = expand_alpha_rev_1x64 (load_32_1x64 (m)); - ms = unpack_32_1x64 (s); - md = unpack_32_1x64 (d); + ma = expand_alpha_rev_1x128 (load_32_1x128 (m)); + ms = unpack_32_1x128 (s); + md = unpack_32_1x128 (d); - msa = expand_alpha_rev_1x64 (load_32_1x64 (sa)); + msa = expand_alpha_rev_1x128 (load_32_1x128 (sa)); - *dst = pack_1x64_32 (in_over_1x64 (&ms, &msa, &ma, &md)); + *dst = pack_1x128_32 (in_over_1x128 (&ms, &msa, &ma, &md)); } } @@ -5546,7 +5107,6 @@ sse2_composite_over_8888_8_8888 (pixman_implementation_t *imp, } } - _mm_empty (); } static void @@ -5591,12 +5151,12 @@ sse2_composite_over_reverse_n_8888 (pixman_implementation_t *imp, while (w && (unsigned long)dst & 15) { - __m64 vd; + __m128i vd; - vd = unpack_32_1x64 (*dst); + vd = unpack_32_1x128 (*dst); - *dst = pack_1x64_32 (over_1x64 (vd, expand_alpha_1x64 (vd), - _mm_movepi64_pi64 (xmm_src))); + *dst = pack_1x128_32 (over_1x128 (vd, expand_alpha_1x128 (vd), + xmm_src)); w--; dst++; } @@ -5626,19 +5186,18 @@ sse2_composite_over_reverse_n_8888 (pixman_implementation_t *imp, while (w) { - __m64 vd; + __m128i vd; - vd = unpack_32_1x64 (*dst); + vd = unpack_32_1x128 (*dst); - *dst = pack_1x64_32 (over_1x64 (vd, expand_alpha_1x64 (vd), - _mm_movepi64_pi64 (xmm_src))); + *dst = pack_1x128_32 (over_1x128 (vd, expand_alpha_1x128 (vd), + xmm_src)); w--; dst++; } } - _mm_empty (); } static void @@ -5703,15 +5262,15 @@ sse2_composite_over_8888_8888_8888 (pixman_implementation_t *imp, } else { - __m64 ms, md, ma, msa; + __m128i ms, md, ma, msa; - ma = expand_alpha_rev_1x64 (load_32_1x64 (m)); - ms = unpack_32_1x64 (s); - md = unpack_32_1x64 (d); + ma = expand_alpha_rev_1x128 (load_32_1x128 (m)); + ms = unpack_32_1x128 (s); + md = unpack_32_1x128 (d); - msa = expand_alpha_rev_1x64 (load_32_1x64 (sa)); + msa = expand_alpha_rev_1x128 (load_32_1x128 (sa)); - *dst = pack_1x64_32 (in_over_1x64 (&ms, &msa, &ma, &md)); + *dst = pack_1x128_32 (in_over_1x128 (&ms, &msa, &ma, &md)); } } @@ -5773,15 +5332,15 @@ sse2_composite_over_8888_8888_8888 (pixman_implementation_t *imp, } else { - __m64 ms, md, ma, msa; + __m128i ms, md, ma, msa; - ma = expand_alpha_rev_1x64 (load_32_1x64 (m)); - ms = unpack_32_1x64 (s); - md = unpack_32_1x64 (d); + ma = expand_alpha_rev_1x128 (load_32_1x128 (m)); + ms = unpack_32_1x128 (s); + md = unpack_32_1x128 (d); - msa = expand_alpha_rev_1x64 (load_32_1x64 (sa)); + msa = expand_alpha_rev_1x128 (load_32_1x128 (sa)); - *dst = pack_1x64_32 (in_over_1x64 (&ms, &msa, &ma, &md)); + *dst = pack_1x128_32 (in_over_1x128 (&ms, &msa, &ma, &md)); } } @@ -5790,10 +5349,9 @@ sse2_composite_over_8888_8888_8888 (pixman_implementation_t *imp, } } - _mm_empty (); } -/* A variant of 'core_combine_over_u_sse2' with minor tweaks */ +/* A variant of 'sse2_combine_over_u' with minor tweaks */ static force_inline void scaled_nearest_scanline_sse2_8888_8888_OVER (uint32_t* pd, const uint32_t* ps, @@ -5885,7 +5443,6 @@ scaled_nearest_scanline_sse2_8888_8888_OVER (uint32_t* pd, w--; } - _mm_empty (); } FAST_NEAREST_MAINLOOP (sse2_8888_8888_cover_OVER, @@ -5927,13 +5484,13 @@ scaled_nearest_scanline_sse2_8888_n_8888_OVER (const uint32_t * mask, { uint32_t d = *dst; - __m64 ms = unpack_32_1x64 (s); - __m64 alpha = expand_alpha_1x64 (ms); - __m64 dest = _mm_movepi64_pi64 (xmm_mask); - __m64 alpha_dst = unpack_32_1x64 (d); + __m128i ms = unpack_32_1x128 (s); + __m128i alpha = expand_alpha_1x128 (ms); + __m128i dest = xmm_mask; + __m128i alpha_dst = unpack_32_1x128 (d); - *dst = pack_1x64_32 ( - in_over_1x64 (&ms, &alpha, &dest, &alpha_dst)); + *dst = pack_1x128_32 ( + in_over_1x128 (&ms, &alpha, &dest, &alpha_dst)); } dst++; w--; @@ -5985,20 +5542,19 @@ scaled_nearest_scanline_sse2_8888_n_8888_OVER (const uint32_t * mask, { uint32_t d = *dst; - __m64 ms = unpack_32_1x64 (s); - __m64 alpha = expand_alpha_1x64 (ms); - __m64 mask = _mm_movepi64_pi64 (xmm_mask); - __m64 dest = unpack_32_1x64 (d); + __m128i ms = unpack_32_1x128 (s); + __m128i alpha = expand_alpha_1x128 (ms); + __m128i mask = xmm_mask; + __m128i dest = unpack_32_1x128 (d); - *dst = pack_1x64_32 ( - in_over_1x64 (&ms, &alpha, &mask, &dest)); + *dst = pack_1x128_32 ( + in_over_1x128 (&ms, &alpha, &mask, &dest)); } dst++; w--; } - _mm_empty (); } FAST_NEAREST_MAINLOOP_COMMON (sse2_8888_n_8888_cover_OVER, @@ -6374,20 +5930,7 @@ _pixman_implementation_create_sse2 (pixman_implementation_t *fallback) mask_ff000000 = create_mask_2x32_128 (0xff000000, 0xff000000); mask_alpha = create_mask_2x32_128 (0x00ff0000, 0x00000000); - /* MMX constants */ - mask_x565_rgb = create_mask_2x32_64 (0x000001f0, 0x003f001f); - mask_x565_unpack = create_mask_2x32_64 (0x00000084, 0x04100840); - - mask_x0080 = create_mask_16_64 (0x0080); - mask_x00ff = create_mask_16_64 (0x00ff); - mask_x0101 = create_mask_16_64 (0x0101); - mask_x_alpha = create_mask_2x32_64 (0x00ff0000, 0x00000000); - - _mm_empty (); - /* Set up function pointers */ - - /* SSE code patch for fbcompose.c */ imp->combine_32[PIXMAN_OP_OVER] = sse2_combine_over_u; imp->combine_32[PIXMAN_OP_OVER_REVERSE] = sse2_combine_over_reverse_u; imp->combine_32[PIXMAN_OP_IN] = sse2_combine_in_u; @@ -6420,5 +5963,3 @@ _pixman_implementation_create_sse2 (pixman_implementation_t *fallback) return imp; } - -#endif /* USE_SSE2 */ diff --git a/xorg-server/dix/eventconvert.c b/xorg-server/dix/eventconvert.c index dff1a5631..dd1ca460b 100644 --- a/xorg-server/dix/eventconvert.c +++ b/xorg-server/dix/eventconvert.c @@ -1,734 +1,746 @@ -/*
- * Copyright © 2009 Red Hat, Inc.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice (including the next
- * paragraph) shall be included in all copies or substantial portions of the
- * Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
- * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
- * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
- * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
- * DEALINGS IN THE SOFTWARE.
- *
- */
-
-/**
- * @file eventconvert.c
- * This file contains event conversion routines from InternalEvent to the
- * matching protocol events.
- */
-
-#ifdef HAVE_DIX_CONFIG_H
-#include <dix-config.h>
-#endif
-
-#include <stdint.h>
-#include <X11/X.h>
-#include <X11/extensions/XIproto.h>
-#include <X11/extensions/XI2proto.h>
-#include <X11/extensions/XI.h>
-#include <X11/extensions/XI2.h>
-
-#include "dix.h"
-#include "inputstr.h"
-#include "misc.h"
-#include "eventstr.h"
-#include "exglobals.h"
-#include "eventconvert.h"
-#include "xiquerydevice.h"
-#include "xkbsrv.h"
-
-
-static int countValuators(DeviceEvent *ev, int *first);
-static int getValuatorEvents(DeviceEvent *ev, deviceValuator *xv);
-static int eventToKeyButtonPointer(DeviceEvent *ev, xEvent **xi, int *count);
-static int eventToDeviceChanged(DeviceChangedEvent *ev, xEvent **dcce);
-static int eventToDeviceEvent(DeviceEvent *ev, xEvent **xi);
-static int eventToRawEvent(RawDeviceEvent *ev, xEvent **xi);
-
-/* Do not use, read comments below */
-BOOL EventIsKeyRepeat(xEvent *event);
-
-/**
- * Hack to allow detectable autorepeat for core and XI1 events.
- * The sequence number is unused until we send to the client and can be
- * misused to store data. More or less, anyway.
- *
- * Do not use this. It may change any time without warning, eat your babies
- * and piss on your cat.
- */
-static void
-EventSetKeyRepeatFlag(xEvent *event, BOOL on)
-{
- event->u.u.sequenceNumber = on;
-}
-
-/**
- * Check if the event was marked as a repeat event before.
- * NOTE: This is a nasty hack and should NOT be used by anyone else but
- * TryClientEvents.
- */
-BOOL
-EventIsKeyRepeat(xEvent *event)
-{
- return !!event->u.u.sequenceNumber;
-}
-
-/**
- * Convert the given event to the respective core event.
- *
- * Return values:
- * Success ... core contains the matching core event.
- * BadValue .. One or more values in the internal event are invalid.
- * BadMatch .. The event has no core equivalent.
- *
- * @param[in] event The event to convert into a core event.
- * @param[in] core The memory location to store the core event at.
- * @return Success or the matching error code.
- */
-int
-EventToCore(InternalEvent *event, xEvent *core)
-{
- switch(event->any.type)
- {
- case ET_Motion:
- {
- DeviceEvent *e = &event->device_event;
- /* Don't create core motion event if neither x nor y are
- * present */
- if (!BitIsOn(e->valuators.mask, 0) &&
- !BitIsOn(e->valuators.mask, 1))
- return BadMatch;
- }
- /* fallthrough */
- case ET_ButtonPress:
- case ET_ButtonRelease:
- case ET_KeyPress:
- case ET_KeyRelease:
- {
- DeviceEvent *e = &event->device_event;
-
- if (e->detail.key > 0xFF)
- return BadMatch;
-
- memset(core, 0, sizeof(xEvent));
- core->u.u.type = e->type - ET_KeyPress + KeyPress;
- core->u.u.detail = e->detail.key & 0xFF;
- core->u.keyButtonPointer.time = e->time;
- core->u.keyButtonPointer.rootX = e->root_x;
- core->u.keyButtonPointer.rootY = e->root_y;
- core->u.keyButtonPointer.state = e->corestate;
- core->u.keyButtonPointer.root = e->root;
- EventSetKeyRepeatFlag(core, (e->type == ET_KeyPress && e->key_repeat));
- }
- break;
- case ET_ProximityIn:
- case ET_ProximityOut:
- case ET_RawKeyPress:
- case ET_RawKeyRelease:
- case ET_RawButtonPress:
- case ET_RawButtonRelease:
- case ET_RawMotion:
- return BadMatch;
- default:
- /* XXX: */
- ErrorF("[dix] EventToCore: Not implemented yet \n");
- return BadImplementation;
- }
- return Success;
-}
-
-/**
- * Convert the given event to the respective XI 1.x event and store it in
- * xi. xi is allocated on demand and must be freed by the caller.
- * count returns the number of events in xi. If count is 1, and the type of
- * xi is GenericEvent, then xi may be larger than 32 bytes.
- *
- * Return values:
- * Success ... core contains the matching core event.
- * BadValue .. One or more values in the internal event are invalid.
- * BadMatch .. The event has no XI equivalent.
- *
- * @param[in] ev The event to convert into an XI 1 event.
- * @param[out] xi Future memory location for the XI event.
- * @param[out] count Number of elements in xi.
- *
- * @return Success or the error code.
- */
-int
-EventToXI(InternalEvent *ev, xEvent **xi, int *count)
-{
- switch (ev->any.type)
- {
- case ET_Motion:
- case ET_ButtonPress:
- case ET_ButtonRelease:
- case ET_KeyPress:
- case ET_KeyRelease:
- case ET_ProximityIn:
- case ET_ProximityOut:
- return eventToKeyButtonPointer(&ev->device_event, xi, count);
- case ET_DeviceChanged:
- case ET_RawKeyPress:
- case ET_RawKeyRelease:
- case ET_RawButtonPress:
- case ET_RawButtonRelease:
- case ET_RawMotion:
- *count = 0;
- *xi = NULL;
- return BadMatch;
- default:
- break;
- }
-
- ErrorF("[dix] EventToXI: Not implemented for %d \n", ev->any.type);
- return BadImplementation;
-}
-
-/**
- * Convert the given event to the respective XI 2.x event and store it in xi.
- * xi is allocated on demand and must be freed by the caller.
- *
- * Return values:
- * Success ... core contains the matching core event.
- * BadValue .. One or more values in the internal event are invalid.
- * BadMatch .. The event has no XI2 equivalent.
- *
- * @param[in] ev The event to convert into an XI2 event
- * @param[out] xi Future memory location for the XI2 event.
- *
- * @return Success or the error code.
- */
-int
-EventToXI2(InternalEvent *ev, xEvent **xi)
-{
- switch (ev->any.type)
- {
- /* Enter/FocusIn are for grabs. We don't need an actual event, since
- * the real events delivered are triggered elsewhere */
- case ET_Enter:
- case ET_FocusIn:
- *xi = NULL;
- return Success;
- case ET_Motion:
- case ET_ButtonPress:
- case ET_ButtonRelease:
- case ET_KeyPress:
- case ET_KeyRelease:
- return eventToDeviceEvent(&ev->device_event, xi);
- case ET_ProximityIn:
- case ET_ProximityOut:
- *xi = NULL;
- return BadMatch;
- case ET_DeviceChanged:
- return eventToDeviceChanged(&ev->changed_event, xi);
- case ET_RawKeyPress:
- case ET_RawKeyRelease:
- case ET_RawButtonPress:
- case ET_RawButtonRelease:
- case ET_RawMotion:
- return eventToRawEvent(&ev->raw_event, xi);
- default:
- break;
- }
-
- ErrorF("[dix] EventToXI2: Not implemented for %d \n", ev->any.type);
- return BadImplementation;
-}
-
-static int
-eventToKeyButtonPointer(DeviceEvent *ev, xEvent **xi, int *count)
-{
- int num_events;
- int first; /* dummy */
- deviceKeyButtonPointer *kbp;
-
- /* Sorry, XI 1.x protocol restrictions. */
- if (ev->detail.button > 0xFF || ev->deviceid >= 0x80)
- {
- *count = 0;
- return Success;
- }
-
- num_events = (countValuators(ev, &first) + 5)/6; /* valuator ev */
- if (num_events <= 0)
- {
- *count = 0;
- return BadMatch;
- }
-
- num_events++; /* the actual event event */
-
- *xi = calloc(num_events, sizeof(xEvent));
- if (!(*xi))
- {
- return BadAlloc;
- }
-
- kbp = (deviceKeyButtonPointer*)(*xi);
- kbp->detail = ev->detail.button;
- kbp->time = ev->time;
- kbp->root = ev->root;
- kbp->root_x = ev->root_x;
- kbp->root_y = ev->root_y;
- kbp->deviceid = ev->deviceid;
- kbp->state = ev->corestate;
- EventSetKeyRepeatFlag((xEvent*)kbp,
- (ev->type == ET_KeyPress && ev->key_repeat));
-
- if (num_events > 1)
- kbp->deviceid |= MORE_EVENTS;
-
- switch(ev->type)
- {
- case ET_Motion: kbp->type = DeviceMotionNotify; break;
- case ET_ButtonPress: kbp->type = DeviceButtonPress; break;
- case ET_ButtonRelease: kbp->type = DeviceButtonRelease; break;
- case ET_KeyPress: kbp->type = DeviceKeyPress; break;
- case ET_KeyRelease: kbp->type = DeviceKeyRelease; break;
- case ET_ProximityIn: kbp->type = ProximityIn; break;
- case ET_ProximityOut: kbp->type = ProximityOut; break;
- default:
- break;
- }
-
- if (num_events > 1)
- {
- getValuatorEvents(ev, (deviceValuator*)(kbp + 1));
- }
-
- *count = num_events;
- return Success;
-}
-
-
-/**
- * Set first to the first valuator in the event ev and return the number of
- * valuators from first to the last set valuator.
- */
-static int
-countValuators(DeviceEvent *ev, int *first)
-{
- int first_valuator = -1, last_valuator = -1, num_valuators = 0;
- int i;
-
- for (i = 0; i < sizeof(ev->valuators.mask) * 8; i++)
- {
- if (BitIsOn(ev->valuators.mask, i))
- {
- if (first_valuator == -1)
- first_valuator = i;
- last_valuator = i;
- }
- }
-
- if (first_valuator != -1)
- {
- num_valuators = last_valuator - first_valuator + 1;
- *first = first_valuator;
- }
-
- return num_valuators;
-}
-
-static int
-getValuatorEvents(DeviceEvent *ev, deviceValuator *xv)
-{
- int i;
- int state = 0;
- int first_valuator, num_valuators;
-
-
- num_valuators = countValuators(ev, &first_valuator);
- if (num_valuators > 0)
- {
- DeviceIntPtr dev = NULL;
- dixLookupDevice(&dev, ev->deviceid, serverClient, DixUseAccess);
- /* State needs to be assembled BEFORE the device is updated. */
- state = (dev && dev->key) ? XkbStateFieldFromRec(&dev->key->xkbInfo->state) : 0;
- state |= (dev && dev->button) ? (dev->button->state) : 0;
- }
-
- /* FIXME: non-continuous valuator data in internal events*/
- for (i = 0; i < num_valuators; i += 6, xv++) {
- xv->type = DeviceValuator;
- xv->first_valuator = first_valuator + i;
- xv->num_valuators = ((num_valuators - i) > 6) ? 6 : (num_valuators - i);
- xv->deviceid = ev->deviceid;
- xv->device_state = state;
- switch (xv->num_valuators) {
- case 6:
- xv->valuator5 = ev->valuators.data[xv->first_valuator + 5];
- case 5:
- xv->valuator4 = ev->valuators.data[xv->first_valuator + 4];
- case 4:
- xv->valuator3 = ev->valuators.data[xv->first_valuator + 3];
- case 3:
- xv->valuator2 = ev->valuators.data[xv->first_valuator + 2];
- case 2:
- xv->valuator1 = ev->valuators.data[xv->first_valuator + 1];
- case 1:
- xv->valuator0 = ev->valuators.data[xv->first_valuator + 0];
- }
-
- if (i + 6 < num_valuators)
- xv->deviceid |= MORE_EVENTS;
- }
-
- return (num_valuators + 5) / 6;
-}
-
-
-static int
-appendKeyInfo(DeviceChangedEvent *dce, xXIKeyInfo* info)
-{
- uint32_t *kc;
- int i;
-
- info->type = XIKeyClass;
- info->num_keycodes = dce->keys.max_keycode - dce->keys.min_keycode + 1;
- info->length = sizeof(xXIKeyInfo)/4 + info->num_keycodes;
- info->sourceid = dce->sourceid;
-
- kc = (uint32_t*)&info[1];
- for (i = 0; i < info->num_keycodes; i++)
- *kc++ = i + dce->keys.min_keycode;
-
- return info->length * 4;
-}
-
-static int
-appendButtonInfo(DeviceChangedEvent *dce, xXIButtonInfo *info)
-{
- unsigned char *bits;
- int mask_len;
-
- mask_len = bytes_to_int32(bits_to_bytes(dce->buttons.num_buttons));
-
- info->type = XIButtonClass;
- info->num_buttons = dce->buttons.num_buttons;
- info->length = bytes_to_int32(sizeof(xXIButtonInfo)) +
- info->num_buttons + mask_len;
- info->sourceid = dce->sourceid;
-
- bits = (unsigned char*)&info[1];
- memset(bits, 0, mask_len * 4);
- /* FIXME: is_down? */
-
- bits += mask_len * 4;
- memcpy(bits, dce->buttons.names, dce->buttons.num_buttons * sizeof(Atom));
-
- return info->length * 4;
-}
-
-static int
-appendValuatorInfo(DeviceChangedEvent *dce, xXIValuatorInfo *info, int axisnumber)
-{
- info->type = XIValuatorClass;
- info->length = sizeof(xXIValuatorInfo)/4;
- info->label = dce->valuators[axisnumber].name;
- info->min.integral = dce->valuators[axisnumber].min;
- info->min.frac = 0;
- info->max.integral = dce->valuators[axisnumber].max;
- info->max.frac = 0;
- /* FIXME: value */
- info->value.integral = 0;
- info->value.frac = 0;
- info->resolution = dce->valuators[axisnumber].resolution;
- info->number = axisnumber;
- info->mode = dce->valuators[axisnumber].mode;
- info->sourceid = dce->sourceid;
-
- return info->length * 4;
-}
-
-static int
-eventToDeviceChanged(DeviceChangedEvent *dce, xEvent **xi)
-{
- xXIDeviceChangedEvent *dcce;
- int len = sizeof(xXIDeviceChangedEvent);
- int nkeys;
- char *ptr;
-
- if (dce->buttons.num_buttons)
- {
- len += sizeof(xXIButtonInfo);
- len += dce->buttons.num_buttons * sizeof(Atom); /* button names */
- len += pad_to_int32(bits_to_bytes(dce->buttons.num_buttons));
- }
- if (dce->num_valuators)
- len += sizeof(xXIValuatorInfo) * dce->num_valuators;
-
- nkeys = (dce->keys.max_keycode > 0) ?
- dce->keys.max_keycode - dce->keys.min_keycode + 1 : 0;
- if (nkeys > 0)
- {
- len += sizeof(xXIKeyInfo);
- len += sizeof(CARD32) * nkeys; /* keycodes */
- }
-
- dcce = calloc(1, len);
- if (!dcce)
- {
- ErrorF("[Xi] BadAlloc in SendDeviceChangedEvent.\n");
- return BadAlloc;
- }
-
- dcce->type = GenericEvent;
- dcce->extension = IReqCode;
- dcce->evtype = XI_DeviceChanged;
- dcce->time = dce->time;
- dcce->deviceid = dce->deviceid;
- dcce->sourceid = dce->sourceid;
- dcce->reason = (dce->flags & DEVCHANGE_DEVICE_CHANGE) ? XIDeviceChange : XISlaveSwitch;
- dcce->num_classes = 0;
- dcce->length = bytes_to_int32(len - sizeof(xEvent));
-
- ptr = (char*)&dcce[1];
- if (dce->buttons.num_buttons)
- {
- dcce->num_classes++;
- ptr += appendButtonInfo(dce, (xXIButtonInfo*)ptr);
- }
-
- if (nkeys)
- {
- dcce->num_classes++;
- ptr += appendKeyInfo(dce, (xXIKeyInfo*)ptr);
- }
-
- if (dce->num_valuators)
- {
- int i;
-
- dcce->num_classes += dce->num_valuators;
- for (i = 0; i < dce->num_valuators; i++)
- ptr += appendValuatorInfo(dce, (xXIValuatorInfo*)ptr, i);
- }
-
- *xi = (xEvent*)dcce;
-
- return Success;
-}
-
-static int count_bits(unsigned char* ptr, int len)
-{
- int bits = 0;
- unsigned int i;
- unsigned char x;
-
- for (i = 0; i < len; i++)
- {
- x = ptr[i];
- while(x > 0)
- {
- bits += (x & 0x1);
- x >>= 1;
- }
- }
- return bits;
-}
-
-static int
-eventToDeviceEvent(DeviceEvent *ev, xEvent **xi)
-{
- int len = sizeof(xXIDeviceEvent);
- xXIDeviceEvent *xde;
- int i, btlen, vallen;
- char *ptr;
- FP3232 *axisval;
-
- /* FIXME: this should just send the buttons we have, not MAX_BUTTONs. Same
- * with MAX_VALUATORS below */
- /* btlen is in 4 byte units */
- btlen = bytes_to_int32(bits_to_bytes(MAX_BUTTONS));
- len += btlen * 4; /* buttonmask len */
-
-
- vallen = count_bits(ev->valuators.mask, sizeof(ev->valuators.mask)/sizeof(ev->valuators.mask[0]));
- len += vallen * 2 * sizeof(uint32_t); /* axisvalues */
- vallen = bytes_to_int32(bits_to_bytes(MAX_VALUATORS));
- len += vallen * 4; /* valuators mask */
-
- *xi = calloc(1, len);
- xde = (xXIDeviceEvent*)*xi;
- xde->type = GenericEvent;
- xde->extension = IReqCode;
- xde->evtype = GetXI2Type((InternalEvent*)ev);
- xde->time = ev->time;
- xde->length = bytes_to_int32(len - sizeof(xEvent));
- xde->detail = ev->detail.button;
- xde->root = ev->root;
- xde->buttons_len = btlen;
- xde->valuators_len = vallen;
- xde->deviceid = ev->deviceid;
- xde->sourceid = ev->sourceid;
- xde->root_x = FP1616(ev->root_x, ev->root_x_frac);
- xde->root_y = FP1616(ev->root_y, ev->root_y_frac);
-
- if (ev->key_repeat)
- xde->flags |= XIKeyRepeat;
-
- xde->mods.base_mods = ev->mods.base;
- xde->mods.latched_mods = ev->mods.latched;
- xde->mods.locked_mods = ev->mods.locked;
- xde->mods.effective_mods = ev->mods.effective;
-
- xde->group.base_group = ev->group.base;
- xde->group.latched_group = ev->group.latched;
- xde->group.locked_group = ev->group.locked;
- xde->group.effective_group = ev->group.effective;
-
- ptr = (char*)&xde[1];
- for (i = 0; i < sizeof(ev->buttons) * 8; i++)
- {
- if (BitIsOn(ev->buttons, i))
- SetBit(ptr, i);
- }
-
- ptr += xde->buttons_len * 4;
- axisval = (FP3232*)(ptr + xde->valuators_len * 4);
- for (i = 0; i < sizeof(ev->valuators.mask) * 8; i++)
- {
- if (BitIsOn(ev->valuators.mask, i))
- {
- SetBit(ptr, i);
- axisval->integral = ev->valuators.data[i];
- axisval->frac = ev->valuators.data_frac[i];
- axisval++;
- }
- }
-
- return Success;
-}
-
-static int
-eventToRawEvent(RawDeviceEvent *ev, xEvent **xi)
-{
- xXIRawEvent* raw;
- int vallen, nvals;
- int i, len = sizeof(xXIRawEvent);
- char *ptr;
- FP3232 *axisval;
-
- nvals = count_bits(ev->valuators.mask, sizeof(ev->valuators.mask));
- len += nvals * sizeof(FP3232) * 2; /* 8 byte per valuator, once
- raw, once processed */
- vallen = bytes_to_int32(bits_to_bytes(MAX_VALUATORS));
- len += vallen * 4; /* valuators mask */
-
- *xi = calloc(1, len);
- raw = (xXIRawEvent*)*xi;
- raw->type = GenericEvent;
- raw->extension = IReqCode;
- raw->evtype = GetXI2Type((InternalEvent*)ev);
- raw->time = ev->time;
- raw->length = bytes_to_int32(len - sizeof(xEvent));
- raw->detail = ev->detail.button;
- raw->deviceid = ev->deviceid;
- raw->valuators_len = vallen;
-
- ptr = (char*)&raw[1];
- axisval = (FP3232*)(ptr + raw->valuators_len * 4);
- for (i = 0; i < sizeof(ev->valuators.mask) * 8; i++)
- {
- if (BitIsOn(ev->valuators.mask, i))
- {
- SetBit(ptr, i);
- axisval->integral = ev->valuators.data[i];
- axisval->frac = ev->valuators.data_frac[i];
- (axisval + nvals)->integral = ev->valuators.data_raw[i];
- (axisval + nvals)->frac = ev->valuators.data_raw_frac[i];
- axisval++;
- }
- }
-
- return Success;
-}
-
-/**
- * Return the corresponding core type for the given event or 0 if no core
- * equivalent exists.
- */
-int
-GetCoreType(InternalEvent *event)
-{
- int coretype = 0;
- switch(event->any.type)
- {
- case ET_Motion: coretype = MotionNotify; break;
- case ET_ButtonPress: coretype = ButtonPress; break;
- case ET_ButtonRelease: coretype = ButtonRelease; break;
- case ET_KeyPress: coretype = KeyPress; break;
- case ET_KeyRelease: coretype = KeyRelease; break;
- default:
- break;
- }
- return coretype;
-}
-
-/**
- * Return the corresponding XI 1.x type for the given event or 0 if no
- * equivalent exists.
- */
-int
-GetXIType(InternalEvent *event)
-{
- int xitype = 0;
- switch(event->any.type)
- {
- case ET_Motion: xitype = DeviceMotionNotify; break;
- case ET_ButtonPress: xitype = DeviceButtonPress; break;
- case ET_ButtonRelease: xitype = DeviceButtonRelease; break;
- case ET_KeyPress: xitype = DeviceKeyPress; break;
- case ET_KeyRelease: xitype = DeviceKeyRelease; break;
- case ET_ProximityIn: xitype = ProximityIn; break;
- case ET_ProximityOut: xitype = ProximityOut; break;
- default:
- break;
- }
- return xitype;
-}
-
-/**
- * Return the corresponding XI 2.x type for the given event or 0 if no
- * equivalent exists.
- */
-int
-GetXI2Type(InternalEvent *event)
-{
- int xi2type = 0;
-
- switch(event->any.type)
- {
- case ET_Motion: xi2type = XI_Motion; break;
- case ET_ButtonPress: xi2type = XI_ButtonPress; break;
- case ET_ButtonRelease: xi2type = XI_ButtonRelease; break;
- case ET_KeyPress: xi2type = XI_KeyPress; break;
- case ET_KeyRelease: xi2type = XI_KeyRelease; break;
- case ET_Enter: xi2type = XI_Enter; break;
- case ET_Leave: xi2type = XI_Leave; break;
- case ET_Hierarchy: xi2type = XI_HierarchyChanged; break;
- case ET_DeviceChanged: xi2type = XI_DeviceChanged; break;
- case ET_RawKeyPress: xi2type = XI_RawKeyPress; break;
- case ET_RawKeyRelease: xi2type = XI_RawKeyRelease; break;
- case ET_RawButtonPress: xi2type = XI_RawButtonPress; break;
- case ET_RawButtonRelease: xi2type = XI_RawButtonRelease; break;
- case ET_RawMotion: xi2type = XI_RawMotion; break;
- case ET_FocusIn: xi2type = XI_FocusIn; break;
- case ET_FocusOut: xi2type = XI_FocusOut; break;
- default:
- break;
- }
- return xi2type;
-}
+/* + * Copyright © 2009 Red Hat, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + */ + +/** + * @file eventconvert.c + * This file contains event conversion routines from InternalEvent to the + * matching protocol events. + */ + +#ifdef HAVE_DIX_CONFIG_H +#include <dix-config.h> +#endif + +#include <stdint.h> +#include <X11/X.h> +#include <X11/extensions/XIproto.h> +#include <X11/extensions/XI2proto.h> +#include <X11/extensions/XI.h> +#include <X11/extensions/XI2.h> + +#include "dix.h" +#include "inputstr.h" +#include "misc.h" +#include "eventstr.h" +#include "exglobals.h" +#include "eventconvert.h" +#include "xiquerydevice.h" +#include "xkbsrv.h" + + +static int countValuators(DeviceEvent *ev, int *first); +static int getValuatorEvents(DeviceEvent *ev, deviceValuator *xv); +static int eventToKeyButtonPointer(DeviceEvent *ev, xEvent **xi, int *count); +static int eventToDeviceChanged(DeviceChangedEvent *ev, xEvent **dcce); +static int eventToDeviceEvent(DeviceEvent *ev, xEvent **xi); +static int eventToRawEvent(RawDeviceEvent *ev, xEvent **xi); + +/* Do not use, read comments below */ +BOOL EventIsKeyRepeat(xEvent *event); + +/** + * Hack to allow detectable autorepeat for core and XI1 events. + * The sequence number is unused until we send to the client and can be + * misused to store data. More or less, anyway. + * + * Do not use this. It may change any time without warning, eat your babies + * and piss on your cat. + */ +static void +EventSetKeyRepeatFlag(xEvent *event, BOOL on) +{ + event->u.u.sequenceNumber = on; +} + +/** + * Check if the event was marked as a repeat event before. + * NOTE: This is a nasty hack and should NOT be used by anyone else but + * TryClientEvents. + */ +BOOL +EventIsKeyRepeat(xEvent *event) +{ + return !!event->u.u.sequenceNumber; +} + +/** + * Convert the given event to the respective core event. + * + * Return values: + * Success ... core contains the matching core event. + * BadValue .. One or more values in the internal event are invalid. + * BadMatch .. The event has no core equivalent. + * + * @param[in] event The event to convert into a core event. + * @param[in] core The memory location to store the core event at. + * @return Success or the matching error code. + */ +int +EventToCore(InternalEvent *event, xEvent *core) +{ + switch(event->any.type) + { + case ET_Motion: + { + DeviceEvent *e = &event->device_event; + /* Don't create core motion event if neither x nor y are + * present */ + if (!BitIsOn(e->valuators.mask, 0) && + !BitIsOn(e->valuators.mask, 1)) + return BadMatch; + } + /* fallthrough */ + case ET_ButtonPress: + case ET_ButtonRelease: + case ET_KeyPress: + case ET_KeyRelease: + { + DeviceEvent *e = &event->device_event; + + if (e->detail.key > 0xFF) + return BadMatch; + + memset(core, 0, sizeof(xEvent)); + core->u.u.type = e->type - ET_KeyPress + KeyPress; + core->u.u.detail = e->detail.key & 0xFF; + core->u.keyButtonPointer.time = e->time; + core->u.keyButtonPointer.rootX = e->root_x; + core->u.keyButtonPointer.rootY = e->root_y; + core->u.keyButtonPointer.state = e->corestate; + core->u.keyButtonPointer.root = e->root; + EventSetKeyRepeatFlag(core, (e->type == ET_KeyPress && e->key_repeat)); + } + break; + case ET_ProximityIn: + case ET_ProximityOut: + case ET_RawKeyPress: + case ET_RawKeyRelease: + case ET_RawButtonPress: + case ET_RawButtonRelease: + case ET_RawMotion: + return BadMatch; + default: + /* XXX: */ + ErrorF("[dix] EventToCore: Not implemented yet \n"); + return BadImplementation; + } + return Success; +} + +/** + * Convert the given event to the respective XI 1.x event and store it in + * xi. xi is allocated on demand and must be freed by the caller. + * count returns the number of events in xi. If count is 1, and the type of + * xi is GenericEvent, then xi may be larger than 32 bytes. + * + * Return values: + * Success ... core contains the matching core event. + * BadValue .. One or more values in the internal event are invalid. + * BadMatch .. The event has no XI equivalent. + * + * @param[in] ev The event to convert into an XI 1 event. + * @param[out] xi Future memory location for the XI event. + * @param[out] count Number of elements in xi. + * + * @return Success or the error code. + */ +int +EventToXI(InternalEvent *ev, xEvent **xi, int *count) +{ + switch (ev->any.type) + { + case ET_Motion: + case ET_ButtonPress: + case ET_ButtonRelease: + case ET_KeyPress: + case ET_KeyRelease: + case ET_ProximityIn: + case ET_ProximityOut: + return eventToKeyButtonPointer(&ev->device_event, xi, count); + case ET_DeviceChanged: + case ET_RawKeyPress: + case ET_RawKeyRelease: + case ET_RawButtonPress: + case ET_RawButtonRelease: + case ET_RawMotion: + *count = 0; + *xi = NULL; + return BadMatch; + default: + break; + } + + ErrorF("[dix] EventToXI: Not implemented for %d \n", ev->any.type); + return BadImplementation; +} + +/** + * Convert the given event to the respective XI 2.x event and store it in xi. + * xi is allocated on demand and must be freed by the caller. + * + * Return values: + * Success ... core contains the matching core event. + * BadValue .. One or more values in the internal event are invalid. + * BadMatch .. The event has no XI2 equivalent. + * + * @param[in] ev The event to convert into an XI2 event + * @param[out] xi Future memory location for the XI2 event. + * + * @return Success or the error code. + */ +int +EventToXI2(InternalEvent *ev, xEvent **xi) +{ + switch (ev->any.type) + { + /* Enter/FocusIn are for grabs. We don't need an actual event, since + * the real events delivered are triggered elsewhere */ + case ET_Enter: + case ET_FocusIn: + *xi = NULL; + return Success; + case ET_Motion: + case ET_ButtonPress: + case ET_ButtonRelease: + case ET_KeyPress: + case ET_KeyRelease: + return eventToDeviceEvent(&ev->device_event, xi); + case ET_ProximityIn: + case ET_ProximityOut: + *xi = NULL; + return BadMatch; + case ET_DeviceChanged: + return eventToDeviceChanged(&ev->changed_event, xi); + case ET_RawKeyPress: + case ET_RawKeyRelease: + case ET_RawButtonPress: + case ET_RawButtonRelease: + case ET_RawMotion: + return eventToRawEvent(&ev->raw_event, xi); + default: + break; + } + + ErrorF("[dix] EventToXI2: Not implemented for %d \n", ev->any.type); + return BadImplementation; +} + +static int +eventToKeyButtonPointer(DeviceEvent *ev, xEvent **xi, int *count) +{ + int num_events; + int first; /* dummy */ + deviceKeyButtonPointer *kbp; + + /* Sorry, XI 1.x protocol restrictions. */ + if (ev->detail.button > 0xFF || ev->deviceid >= 0x80) + { + *count = 0; + return Success; + } + + num_events = (countValuators(ev, &first) + 5)/6; /* valuator ev */ + if (num_events <= 0) + { + switch (ev->type) + { + case ET_KeyPress: + case ET_KeyRelease: + case ET_ButtonPress: + case ET_ButtonRelease: + /* no axes is ok */ + break; + case ET_Motion: + case ET_ProximityIn: + case ET_ProximityOut: + *count = 0; + return BadMatch; + } + } + + num_events++; /* the actual event event */ + + *xi = calloc(num_events, sizeof(xEvent)); + if (!(*xi)) + { + return BadAlloc; + } + + kbp = (deviceKeyButtonPointer*)(*xi); + kbp->detail = ev->detail.button; + kbp->time = ev->time; + kbp->root = ev->root; + kbp->root_x = ev->root_x; + kbp->root_y = ev->root_y; + kbp->deviceid = ev->deviceid; + kbp->state = ev->corestate; + EventSetKeyRepeatFlag((xEvent*)kbp, + (ev->type == ET_KeyPress && ev->key_repeat)); + + if (num_events > 1) + kbp->deviceid |= MORE_EVENTS; + + switch(ev->type) + { + case ET_Motion: kbp->type = DeviceMotionNotify; break; + case ET_ButtonPress: kbp->type = DeviceButtonPress; break; + case ET_ButtonRelease: kbp->type = DeviceButtonRelease; break; + case ET_KeyPress: kbp->type = DeviceKeyPress; break; + case ET_KeyRelease: kbp->type = DeviceKeyRelease; break; + case ET_ProximityIn: kbp->type = ProximityIn; break; + case ET_ProximityOut: kbp->type = ProximityOut; break; + default: + break; + } + + if (num_events > 1) + { + getValuatorEvents(ev, (deviceValuator*)(kbp + 1)); + } + + *count = num_events; + return Success; +} + + +/** + * Set first to the first valuator in the event ev and return the number of + * valuators from first to the last set valuator. + */ +static int +countValuators(DeviceEvent *ev, int *first) +{ + int first_valuator = -1, last_valuator = -1, num_valuators = 0; + int i; + + for (i = 0; i < sizeof(ev->valuators.mask) * 8; i++) + { + if (BitIsOn(ev->valuators.mask, i)) + { + if (first_valuator == -1) + first_valuator = i; + last_valuator = i; + } + } + + if (first_valuator != -1) + { + num_valuators = last_valuator - first_valuator + 1; + *first = first_valuator; + } + + return num_valuators; +} + +static int +getValuatorEvents(DeviceEvent *ev, deviceValuator *xv) +{ + int i; + int state = 0; + int first_valuator, num_valuators; + + + num_valuators = countValuators(ev, &first_valuator); + if (num_valuators > 0) + { + DeviceIntPtr dev = NULL; + dixLookupDevice(&dev, ev->deviceid, serverClient, DixUseAccess); + /* State needs to be assembled BEFORE the device is updated. */ + state = (dev && dev->key) ? XkbStateFieldFromRec(&dev->key->xkbInfo->state) : 0; + state |= (dev && dev->button) ? (dev->button->state) : 0; + } + + /* FIXME: non-continuous valuator data in internal events*/ + for (i = 0; i < num_valuators; i += 6, xv++) { + xv->type = DeviceValuator; + xv->first_valuator = first_valuator + i; + xv->num_valuators = ((num_valuators - i) > 6) ? 6 : (num_valuators - i); + xv->deviceid = ev->deviceid; + xv->device_state = state; + switch (xv->num_valuators) { + case 6: + xv->valuator5 = ev->valuators.data[xv->first_valuator + 5]; + case 5: + xv->valuator4 = ev->valuators.data[xv->first_valuator + 4]; + case 4: + xv->valuator3 = ev->valuators.data[xv->first_valuator + 3]; + case 3: + xv->valuator2 = ev->valuators.data[xv->first_valuator + 2]; + case 2: + xv->valuator1 = ev->valuators.data[xv->first_valuator + 1]; + case 1: + xv->valuator0 = ev->valuators.data[xv->first_valuator + 0]; + } + + if (i + 6 < num_valuators) + xv->deviceid |= MORE_EVENTS; + } + + return (num_valuators + 5) / 6; +} + + +static int +appendKeyInfo(DeviceChangedEvent *dce, xXIKeyInfo* info) +{ + uint32_t *kc; + int i; + + info->type = XIKeyClass; + info->num_keycodes = dce->keys.max_keycode - dce->keys.min_keycode + 1; + info->length = sizeof(xXIKeyInfo)/4 + info->num_keycodes; + info->sourceid = dce->sourceid; + + kc = (uint32_t*)&info[1]; + for (i = 0; i < info->num_keycodes; i++) + *kc++ = i + dce->keys.min_keycode; + + return info->length * 4; +} + +static int +appendButtonInfo(DeviceChangedEvent *dce, xXIButtonInfo *info) +{ + unsigned char *bits; + int mask_len; + + mask_len = bytes_to_int32(bits_to_bytes(dce->buttons.num_buttons)); + + info->type = XIButtonClass; + info->num_buttons = dce->buttons.num_buttons; + info->length = bytes_to_int32(sizeof(xXIButtonInfo)) + + info->num_buttons + mask_len; + info->sourceid = dce->sourceid; + + bits = (unsigned char*)&info[1]; + memset(bits, 0, mask_len * 4); + /* FIXME: is_down? */ + + bits += mask_len * 4; + memcpy(bits, dce->buttons.names, dce->buttons.num_buttons * sizeof(Atom)); + + return info->length * 4; +} + +static int +appendValuatorInfo(DeviceChangedEvent *dce, xXIValuatorInfo *info, int axisnumber) +{ + info->type = XIValuatorClass; + info->length = sizeof(xXIValuatorInfo)/4; + info->label = dce->valuators[axisnumber].name; + info->min.integral = dce->valuators[axisnumber].min; + info->min.frac = 0; + info->max.integral = dce->valuators[axisnumber].max; + info->max.frac = 0; + /* FIXME: value */ + info->value.integral = 0; + info->value.frac = 0; + info->resolution = dce->valuators[axisnumber].resolution; + info->number = axisnumber; + info->mode = dce->valuators[axisnumber].mode; + info->sourceid = dce->sourceid; + + return info->length * 4; +} + +static int +eventToDeviceChanged(DeviceChangedEvent *dce, xEvent **xi) +{ + xXIDeviceChangedEvent *dcce; + int len = sizeof(xXIDeviceChangedEvent); + int nkeys; + char *ptr; + + if (dce->buttons.num_buttons) + { + len += sizeof(xXIButtonInfo); + len += dce->buttons.num_buttons * sizeof(Atom); /* button names */ + len += pad_to_int32(bits_to_bytes(dce->buttons.num_buttons)); + } + if (dce->num_valuators) + len += sizeof(xXIValuatorInfo) * dce->num_valuators; + + nkeys = (dce->keys.max_keycode > 0) ? + dce->keys.max_keycode - dce->keys.min_keycode + 1 : 0; + if (nkeys > 0) + { + len += sizeof(xXIKeyInfo); + len += sizeof(CARD32) * nkeys; /* keycodes */ + } + + dcce = calloc(1, len); + if (!dcce) + { + ErrorF("[Xi] BadAlloc in SendDeviceChangedEvent.\n"); + return BadAlloc; + } + + dcce->type = GenericEvent; + dcce->extension = IReqCode; + dcce->evtype = XI_DeviceChanged; + dcce->time = dce->time; + dcce->deviceid = dce->deviceid; + dcce->sourceid = dce->sourceid; + dcce->reason = (dce->flags & DEVCHANGE_DEVICE_CHANGE) ? XIDeviceChange : XISlaveSwitch; + dcce->num_classes = 0; + dcce->length = bytes_to_int32(len - sizeof(xEvent)); + + ptr = (char*)&dcce[1]; + if (dce->buttons.num_buttons) + { + dcce->num_classes++; + ptr += appendButtonInfo(dce, (xXIButtonInfo*)ptr); + } + + if (nkeys) + { + dcce->num_classes++; + ptr += appendKeyInfo(dce, (xXIKeyInfo*)ptr); + } + + if (dce->num_valuators) + { + int i; + + dcce->num_classes += dce->num_valuators; + for (i = 0; i < dce->num_valuators; i++) + ptr += appendValuatorInfo(dce, (xXIValuatorInfo*)ptr, i); + } + + *xi = (xEvent*)dcce; + + return Success; +} + +static int count_bits(unsigned char* ptr, int len) +{ + int bits = 0; + unsigned int i; + unsigned char x; + + for (i = 0; i < len; i++) + { + x = ptr[i]; + while(x > 0) + { + bits += (x & 0x1); + x >>= 1; + } + } + return bits; +} + +static int +eventToDeviceEvent(DeviceEvent *ev, xEvent **xi) +{ + int len = sizeof(xXIDeviceEvent); + xXIDeviceEvent *xde; + int i, btlen, vallen; + char *ptr; + FP3232 *axisval; + + /* FIXME: this should just send the buttons we have, not MAX_BUTTONs. Same + * with MAX_VALUATORS below */ + /* btlen is in 4 byte units */ + btlen = bytes_to_int32(bits_to_bytes(MAX_BUTTONS)); + len += btlen * 4; /* buttonmask len */ + + + vallen = count_bits(ev->valuators.mask, sizeof(ev->valuators.mask)/sizeof(ev->valuators.mask[0])); + len += vallen * 2 * sizeof(uint32_t); /* axisvalues */ + vallen = bytes_to_int32(bits_to_bytes(MAX_VALUATORS)); + len += vallen * 4; /* valuators mask */ + + *xi = calloc(1, len); + xde = (xXIDeviceEvent*)*xi; + xde->type = GenericEvent; + xde->extension = IReqCode; + xde->evtype = GetXI2Type((InternalEvent*)ev); + xde->time = ev->time; + xde->length = bytes_to_int32(len - sizeof(xEvent)); + xde->detail = ev->detail.button; + xde->root = ev->root; + xde->buttons_len = btlen; + xde->valuators_len = vallen; + xde->deviceid = ev->deviceid; + xde->sourceid = ev->sourceid; + xde->root_x = FP1616(ev->root_x, ev->root_x_frac); + xde->root_y = FP1616(ev->root_y, ev->root_y_frac); + + if (ev->key_repeat) + xde->flags |= XIKeyRepeat; + + xde->mods.base_mods = ev->mods.base; + xde->mods.latched_mods = ev->mods.latched; + xde->mods.locked_mods = ev->mods.locked; + xde->mods.effective_mods = ev->mods.effective; + + xde->group.base_group = ev->group.base; + xde->group.latched_group = ev->group.latched; + xde->group.locked_group = ev->group.locked; + xde->group.effective_group = ev->group.effective; + + ptr = (char*)&xde[1]; + for (i = 0; i < sizeof(ev->buttons) * 8; i++) + { + if (BitIsOn(ev->buttons, i)) + SetBit(ptr, i); + } + + ptr += xde->buttons_len * 4; + axisval = (FP3232*)(ptr + xde->valuators_len * 4); + for (i = 0; i < sizeof(ev->valuators.mask) * 8; i++) + { + if (BitIsOn(ev->valuators.mask, i)) + { + SetBit(ptr, i); + axisval->integral = ev->valuators.data[i]; + axisval->frac = ev->valuators.data_frac[i]; + axisval++; + } + } + + return Success; +} + +static int +eventToRawEvent(RawDeviceEvent *ev, xEvent **xi) +{ + xXIRawEvent* raw; + int vallen, nvals; + int i, len = sizeof(xXIRawEvent); + char *ptr; + FP3232 *axisval; + + nvals = count_bits(ev->valuators.mask, sizeof(ev->valuators.mask)); + len += nvals * sizeof(FP3232) * 2; /* 8 byte per valuator, once + raw, once processed */ + vallen = bytes_to_int32(bits_to_bytes(MAX_VALUATORS)); + len += vallen * 4; /* valuators mask */ + + *xi = calloc(1, len); + raw = (xXIRawEvent*)*xi; + raw->type = GenericEvent; + raw->extension = IReqCode; + raw->evtype = GetXI2Type((InternalEvent*)ev); + raw->time = ev->time; + raw->length = bytes_to_int32(len - sizeof(xEvent)); + raw->detail = ev->detail.button; + raw->deviceid = ev->deviceid; + raw->valuators_len = vallen; + + ptr = (char*)&raw[1]; + axisval = (FP3232*)(ptr + raw->valuators_len * 4); + for (i = 0; i < sizeof(ev->valuators.mask) * 8; i++) + { + if (BitIsOn(ev->valuators.mask, i)) + { + SetBit(ptr, i); + axisval->integral = ev->valuators.data[i]; + axisval->frac = ev->valuators.data_frac[i]; + (axisval + nvals)->integral = ev->valuators.data_raw[i]; + (axisval + nvals)->frac = ev->valuators.data_raw_frac[i]; + axisval++; + } + } + + return Success; +} + +/** + * Return the corresponding core type for the given event or 0 if no core + * equivalent exists. + */ +int +GetCoreType(InternalEvent *event) +{ + int coretype = 0; + switch(event->any.type) + { + case ET_Motion: coretype = MotionNotify; break; + case ET_ButtonPress: coretype = ButtonPress; break; + case ET_ButtonRelease: coretype = ButtonRelease; break; + case ET_KeyPress: coretype = KeyPress; break; + case ET_KeyRelease: coretype = KeyRelease; break; + default: + break; + } + return coretype; +} + +/** + * Return the corresponding XI 1.x type for the given event or 0 if no + * equivalent exists. + */ +int +GetXIType(InternalEvent *event) +{ + int xitype = 0; + switch(event->any.type) + { + case ET_Motion: xitype = DeviceMotionNotify; break; + case ET_ButtonPress: xitype = DeviceButtonPress; break; + case ET_ButtonRelease: xitype = DeviceButtonRelease; break; + case ET_KeyPress: xitype = DeviceKeyPress; break; + case ET_KeyRelease: xitype = DeviceKeyRelease; break; + case ET_ProximityIn: xitype = ProximityIn; break; + case ET_ProximityOut: xitype = ProximityOut; break; + default: + break; + } + return xitype; +} + +/** + * Return the corresponding XI 2.x type for the given event or 0 if no + * equivalent exists. + */ +int +GetXI2Type(InternalEvent *event) +{ + int xi2type = 0; + + switch(event->any.type) + { + case ET_Motion: xi2type = XI_Motion; break; + case ET_ButtonPress: xi2type = XI_ButtonPress; break; + case ET_ButtonRelease: xi2type = XI_ButtonRelease; break; + case ET_KeyPress: xi2type = XI_KeyPress; break; + case ET_KeyRelease: xi2type = XI_KeyRelease; break; + case ET_Enter: xi2type = XI_Enter; break; + case ET_Leave: xi2type = XI_Leave; break; + case ET_Hierarchy: xi2type = XI_HierarchyChanged; break; + case ET_DeviceChanged: xi2type = XI_DeviceChanged; break; + case ET_RawKeyPress: xi2type = XI_RawKeyPress; break; + case ET_RawKeyRelease: xi2type = XI_RawKeyRelease; break; + case ET_RawButtonPress: xi2type = XI_RawButtonPress; break; + case ET_RawButtonRelease: xi2type = XI_RawButtonRelease; break; + case ET_RawMotion: xi2type = XI_RawMotion; break; + case ET_FocusIn: xi2type = XI_FocusIn; break; + case ET_FocusOut: xi2type = XI_FocusOut; break; + default: + break; + } + return xi2type; +} diff --git a/xorg-server/test/input.c b/xorg-server/test/input.c index 879e14f2f..e0e9e6af7 100644 --- a/xorg-server/test/input.c +++ b/xorg-server/test/input.c @@ -36,6 +36,7 @@ #include "inputstr.h" #include "eventconvert.h" #include "exevents.h" +#include "exglobals.h" #include "dixgrabs.h" #include "eventstr.h" #include "inpututils.h" @@ -285,6 +286,143 @@ static void dix_event_to_core_conversion(void) dix_event_to_core(ET_Motion); } +static void +_dix_test_xi_convert(DeviceEvent *ev, int expected_rc, int expected_count) +{ + xEvent *xi; + int count = 0; + int rc; + + rc = EventToXI((InternalEvent*)ev, &xi, &count); + g_assert(rc == expected_rc); + g_assert(count >= expected_count); + if (count > 0){ + deviceKeyButtonPointer *kbp = (deviceKeyButtonPointer*)xi; + g_assert(kbp->type == IEventBase + ev->type); + g_assert(kbp->detail == ev->detail.key); + g_assert(kbp->time == ev->time); + g_assert((kbp->deviceid & ~MORE_EVENTS) == ev->deviceid); + g_assert(kbp->root_x == ev->root_x); + g_assert(kbp->root_y == ev->root_y); + g_assert(kbp->state == ev->corestate); + g_assert(kbp->event_x == 0); + g_assert(kbp->event_y == 0); + g_assert(kbp->root == ev->root); + g_assert(kbp->event == 0); + g_assert(kbp->child == 0); + g_assert(kbp->same_screen == FALSE); + + while (--count > 0) { + deviceValuator *v = (deviceValuator*)&xi[count]; + g_assert(v->type == DeviceValuator); + g_assert(v->num_valuators <= 6); + } + + + free(xi); + } +} + +/** + * This tests for internal event → XI1 event conversion + * - all conversions should generate the right XI event type + * - right number of events generated + * - extra events are valuators + */ +static void dix_event_to_xi1_conversion(void) +{ + DeviceEvent ev = {0}; + int time; + int x, y; + int state; + int detail; + const int ROOT_WINDOW_ID = 0x100; + int deviceid; + + IEventBase = 80; + DeviceValuator = IEventBase - 1; + DeviceKeyPress = IEventBase + ET_KeyPress; + DeviceKeyRelease = IEventBase + ET_KeyRelease; + DeviceButtonPress = IEventBase + ET_ButtonPress; + DeviceButtonRelease = IEventBase + ET_ButtonRelease; + DeviceMotionNotify = IEventBase + ET_Motion; + DeviceFocusIn = IEventBase + ET_FocusIn; + DeviceFocusOut = IEventBase + ET_FocusOut; + ProximityIn = IEventBase + ET_ProximityIn; + ProximityOut = IEventBase + ET_ProximityOut; + + /* EventToXI callocs */ + x = 0; + y = 0; + time = 12345; + state = 0; + detail = 0; + deviceid = 4; + + ev.header = 0xFF; + + ev.header = 0xFF; + ev.length = sizeof(DeviceEvent); + ev.time = time; + ev.root_y = x; + ev.root_x = y; + SetBit(ev.valuators.mask, 0); + SetBit(ev.valuators.mask, 1); + ev.root = ROOT_WINDOW_ID; + ev.corestate = state; + ev.detail.key = detail; + ev.deviceid = deviceid; + + /* test all types for bad match */ + ev.type = ET_KeyPress; _dix_test_xi_convert(&ev, Success, 1); + ev.type = ET_KeyRelease; _dix_test_xi_convert(&ev, Success, 1); + ev.type = ET_ButtonPress; _dix_test_xi_convert(&ev, Success, 1); + ev.type = ET_ButtonRelease; _dix_test_xi_convert(&ev, Success, 1); + ev.type = ET_Motion; _dix_test_xi_convert(&ev, Success, 1); + ev.type = ET_ProximityIn; _dix_test_xi_convert(&ev, Success, 1); + ev.type = ET_ProximityOut; _dix_test_xi_convert(&ev, Success, 1); + + /* No axes */ + ClearBit(ev.valuators.mask, 0); + ClearBit(ev.valuators.mask, 1); + ev.type = ET_KeyPress; _dix_test_xi_convert(&ev, Success, 1); + ev.type = ET_KeyRelease; _dix_test_xi_convert(&ev, Success, 1); + ev.type = ET_ButtonPress; _dix_test_xi_convert(&ev, Success, 1); + ev.type = ET_ButtonRelease; _dix_test_xi_convert(&ev, Success, 1); + ev.type = ET_Motion; _dix_test_xi_convert(&ev, BadMatch, 0); + ev.type = ET_ProximityIn; _dix_test_xi_convert(&ev, BadMatch, 0); + ev.type = ET_ProximityOut; _dix_test_xi_convert(&ev, BadMatch, 0); + + /* more than 6 axes → 2 valuator events */ + SetBit(ev.valuators.mask, 0); + SetBit(ev.valuators.mask, 1); + SetBit(ev.valuators.mask, 2); + SetBit(ev.valuators.mask, 3); + SetBit(ev.valuators.mask, 4); + SetBit(ev.valuators.mask, 5); + SetBit(ev.valuators.mask, 6); + ev.type = ET_KeyPress; _dix_test_xi_convert(&ev, Success, 2); + ev.type = ET_KeyRelease; _dix_test_xi_convert(&ev, Success, 2); + ev.type = ET_ButtonPress; _dix_test_xi_convert(&ev, Success, 2); + ev.type = ET_ButtonRelease; _dix_test_xi_convert(&ev, Success, 2); + ev.type = ET_Motion; _dix_test_xi_convert(&ev, Success, 2); + ev.type = ET_ProximityIn; _dix_test_xi_convert(&ev, Success, 2); + ev.type = ET_ProximityOut; _dix_test_xi_convert(&ev, Success, 2); + + + /* keycode too high */ + ev.type = ET_KeyPress; + ev.detail.key = 256; + _dix_test_xi_convert(&ev, Success, 0); + + /* deviceid too high */ + ev.type = ET_KeyPress; + ev.detail.key = 18; + ev.deviceid = 128; + _dix_test_xi_convert(&ev, Success, 0); +} + + static void xi2_struct_sizes(void) { #define compare(req) \ @@ -1070,6 +1208,7 @@ int main(int argc, char** argv) g_test_add_func("/dix/input/attributes", dix_input_attributes); g_test_add_func("/dix/input/init-valuators", dix_init_valuators); g_test_add_func("/dix/input/event-core-conversion", dix_event_to_core_conversion); + g_test_add_func("/dix/input/event-xi1-conversion", dix_event_to_xi1_conversion); g_test_add_func("/dix/input/check-grab-values", dix_check_grab_values); g_test_add_func("/dix/input/xi2-struct-sizes", xi2_struct_sizes); g_test_add_func("/dix/input/grab_matching", dix_grab_matching); |