/*
 * (C) Copyright IBM Corporation 2004, 2005
 * 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
 * IBM,
 * 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.
 */

#include <inttypes.h>
#include <assert.h>
#include <string.h>

#include "glxclient.h"
#include "indirect.h"
#include <GL/glxproto.h>
#include "glxextensions.h"
#include "indirect_vertex_array.h"
#include "indirect_vertex_array_priv.h"

#define __GLX_PAD(n) (((n)+3) & ~3)

/**
 * \file indirect_vertex_array.c
 * Implement GLX protocol for vertex arrays and vertex buffer objects.
 * 
 * The most important function in this fill is \c fill_array_info_cache.
 * The \c array_state_vector contains a cache of the ARRAY_INFO data sent
 * in the DrawArrays protocol.  Certain operations, such as enabling or
 * disabling an array, can invalidate this cache.  \c fill_array_info_cache
 * fills-in this data.  Additionally, it examines the enabled state and
 * other factors to determine what "version" of DrawArrays protocoal can be
 * used.
 * 
 * Current, only two versions of DrawArrays protocol are implemented.  The
 * first version is the "none" protocol.  This is the fallback when the
 * server does not support GL 1.1 / EXT_vertex_arrays.  It is implemented
 * by sending batches of immediate mode commands that are equivalent to the
 * DrawArrays protocol.
 *
 * The other protocol that is currently implemented is the "old" protocol.
 * This is the GL 1.1 DrawArrays protocol.  The only difference between GL
 * 1.1 and EXT_vertex_arrays is the opcode used for the DrawArrays command.
 * This protocol is called "old" because the ARB is in the process of
 * defining a new protocol, which will probably be called wither "new" or
 * "vbo", to support multiple texture coordinate arrays, generic attributes,
 * and vertex buffer objects.
 *
 * \author Ian Romanick <ian.d.romanick@intel.com>
 */

static void emit_DrawArrays_none( GLenum mode, GLint first, GLsizei count );
static void emit_DrawArrays_old ( GLenum mode, GLint first, GLsizei count );

static void emit_DrawElements_none( GLenum mode, GLsizei count, GLenum type,
    const GLvoid *indices );
static void emit_DrawElements_old ( GLenum mode, GLsizei count, GLenum type,
    const GLvoid *indices );


static GLubyte * emit_element_none( GLubyte * dst,
    const struct array_state_vector * arrays, unsigned index );
static GLubyte * emit_element_old( GLubyte * dst,
    const struct array_state_vector * arrays, unsigned index );
static struct array_state * get_array_entry(
    const struct array_state_vector * arrays, GLenum key, unsigned index );
static void fill_array_info_cache( struct array_state_vector * arrays );
static GLboolean validate_mode(__GLXcontext *gc, GLenum mode);
static GLboolean validate_count(__GLXcontext *gc, GLsizei count);
static GLboolean validate_type(__GLXcontext *gc, GLenum type);


/**
 * Table of sizes, in bytes, of a GL types.  All of the type enums are be in 
 * the range 0x1400 - 0x140F.  That includes types added by extensions (i.e.,
 * \c GL_HALF_FLOAT_NV).  This elements of this table correspond to the
 * type enums masked with 0x0f.
 * 
 * \notes
 * \c GL_HALF_FLOAT_NV is not included.  Neither are \c GL_2_BYTES,
 * \c GL_3_BYTES, or \c GL_4_BYTES.
 */
const GLuint __glXTypeSize_table[16] = {
    1, 1, 2, 2, 4, 4, 4, 0, 0, 0, 8, 0, 0, 0, 0, 0
};


/**
 * Free the per-context array state that was allocated with
 * __glXInitVertexArrayState().
 */
void
__glXFreeVertexArrayState( __GLXcontext * gc )
{
    __GLXattribute * state = (__GLXattribute *)(gc->client_state_private);
    struct array_state_vector* arrays = state->array_state;

    if (arrays) {
        if (arrays->stack) {
            free(arrays->stack);
            arrays->stack = NULL;
        }
        if (arrays->arrays) {
            free(arrays->arrays);
            arrays->arrays = NULL;
        }
        free(arrays);
        arrays = NULL;
        state->array_state = NULL;
    }
}


/**
 * Initialize vertex array state of a GLX context.
 * 
 * \param gc  GLX context whose vertex array state is to be initialized.
 * 
 * \warning
 * This function may only be called after __GLXcontext::gl_extension_bits,
 * __GLXcontext::server_minor, and __GLXcontext::server_major have been
 * initialized.  These values are used to determine what vertex arrays are
 * supported.
 * 
 * \bug
 * Return values from malloc are not properly tested.
 */
void
__glXInitVertexArrayState( __GLXcontext * gc )
{
    __GLXattribute * state = (__GLXattribute *)(gc->client_state_private);
    struct array_state_vector * arrays;

    unsigned array_count;
    int texture_units = 1, vertex_program_attribs = 0;
    unsigned i, j;

    GLboolean got_fog = GL_FALSE;
    GLboolean got_secondary_color = GL_FALSE;


    arrays = calloc( 1, sizeof( struct array_state_vector ) );
    state->array_state = arrays;

    arrays->old_DrawArrays_possible = !state->NoDrawArraysProtocol;
    arrays->new_DrawArrays_possible = GL_FALSE;
    arrays->DrawArrays = NULL;

    arrays->active_texture_unit = 0;


    /* Determine how many arrays are actually needed.  Only arrays that
     * are supported by the server are create.  For example, if the server
     * supports only 2 texture units, then only 2 texture coordinate arrays
     * are created.
     * 
     * At the very least, GL_VERTEX_ARRAY, GL_NORMAL_ARRAY,
     * GL_COLOR_ARRAY, GL_INDEX_ARRAY, GL_TEXTURE_COORD_ARRAY, and
     * GL_EDGE_FLAG_ARRAY are supported.
     */

    array_count = 5;
    
    if ( __glExtensionBitIsEnabled( gc, GL_EXT_fog_coord_bit )
	 || (gc->server_major > 1) || (gc->server_minor >= 4) ) {
	got_fog = GL_TRUE;
	array_count++;
    }

    if ( __glExtensionBitIsEnabled( gc, GL_EXT_secondary_color_bit )
	 || (gc->server_major > 1) || (gc->server_minor >= 4) ) {
	got_secondary_color = GL_TRUE;
	array_count++;
    }

    if ( __glExtensionBitIsEnabled( gc, GL_ARB_multitexture_bit )
	 || (gc->server_major > 1) || (gc->server_minor >= 3) ) {
	__indirect_glGetIntegerv( GL_MAX_TEXTURE_UNITS, & texture_units );
    }

    if ( __glExtensionBitIsEnabled( gc, GL_ARB_vertex_program_bit ) ) {
	__indirect_glGetProgramivARB( GL_VERTEX_PROGRAM_ARB,
				      GL_MAX_PROGRAM_ATTRIBS_ARB,
				      & vertex_program_attribs );
    }

    arrays->num_texture_units = texture_units;
    arrays->num_vertex_program_attribs = vertex_program_attribs;
    array_count += texture_units + vertex_program_attribs;
    arrays->num_arrays = array_count;
    arrays->arrays = calloc( array_count, sizeof( struct array_state ) );

    arrays->arrays[0].data_type = GL_FLOAT;
    arrays->arrays[0].count = 3;
    arrays->arrays[0].key = GL_NORMAL_ARRAY;
    arrays->arrays[0].normalized = GL_TRUE;
    arrays->arrays[0].old_DrawArrays_possible = GL_TRUE;

    arrays->arrays[1].data_type = GL_FLOAT;
    arrays->arrays[1].count = 4;
    arrays->arrays[1].key = GL_COLOR_ARRAY;
    arrays->arrays[1].normalized = GL_TRUE;
    arrays->arrays[1].old_DrawArrays_possible = GL_TRUE;

    arrays->arrays[2].data_type = GL_FLOAT;
    arrays->arrays[2].count = 1;
    arrays->arrays[2].key = GL_INDEX_ARRAY;
    arrays->arrays[2].old_DrawArrays_possible = GL_TRUE;

    arrays->arrays[3].data_type = GL_UNSIGNED_BYTE;
    arrays->arrays[3].count = 1;
    arrays->arrays[3].key = GL_EDGE_FLAG_ARRAY;
    arrays->arrays[3].old_DrawArrays_possible = GL_TRUE;

    for ( i = 0 ; i < texture_units ; i++ ) {
	arrays->arrays[4 + i].data_type = GL_FLOAT;
	arrays->arrays[4 + i].count = 4;
	arrays->arrays[4 + i].key = GL_TEXTURE_COORD_ARRAY;

	arrays->arrays[4 + i].old_DrawArrays_possible = (i == 0);
	arrays->arrays[4 + i].index = i;

	arrays->arrays[4 + i].header[1] = i + GL_TEXTURE0;
    }
    
    i = 4 + texture_units;

    if ( got_fog ) {
	arrays->arrays[i].data_type = GL_FLOAT;
	arrays->arrays[i].count = 1;
	arrays->arrays[i].key = GL_FOG_COORDINATE_ARRAY;
	arrays->arrays[i].old_DrawArrays_possible = GL_TRUE;
	i++;
    }

    if ( got_secondary_color ) {
	arrays->arrays[i].data_type = GL_FLOAT;
	arrays->arrays[i].count = 3;
	arrays->arrays[i].key = GL_SECONDARY_COLOR_ARRAY;
	arrays->arrays[i].old_DrawArrays_possible = GL_TRUE;
	arrays->arrays[i].normalized = GL_TRUE;
	i++;
    }


    for ( j = 0 ; j < vertex_program_attribs ; j++ ) {
	const unsigned idx = (vertex_program_attribs - (j + 1));


	arrays->arrays[idx + i].data_type = GL_FLOAT;
	arrays->arrays[idx + i].count = 4;
	arrays->arrays[idx + i].key = GL_VERTEX_ATTRIB_ARRAY_POINTER;

	arrays->arrays[idx + i].old_DrawArrays_possible = 0;
	arrays->arrays[idx + i].index = idx;

	arrays->arrays[idx + i].header[1] = idx;
    }

    i += vertex_program_attribs;


    /* Vertex array *must* be last becuase of the way that
     * emit_DrawArrays_none works.
     */

    arrays->arrays[i].data_type = GL_FLOAT;
    arrays->arrays[i].count = 4;
    arrays->arrays[i].key = GL_VERTEX_ARRAY;
    arrays->arrays[i].old_DrawArrays_possible = GL_TRUE;

    assert( (i + 1) == arrays->num_arrays );

    arrays->stack_index = 0;
    arrays->stack = malloc( sizeof( struct array_stack_state )
			    * arrays->num_arrays );
}


/**
 * Calculate the size of a single vertex for the "none" protocol.  This is
 * essentially the size of all the immediate-mode commands required to
 * implement the enabled vertex arrays.
 */
static size_t
calculate_single_vertex_size_none( const struct array_state_vector * arrays )
{
    size_t single_vertex_size = 0;
    unsigned   i;


    for ( i = 0 ; i < arrays->num_arrays ; i++ ) {
    	if ( arrays->arrays[i].enabled ) {
	    single_vertex_size += ((uint16_t *)arrays->arrays[i].header)[0];
	}
    }
    
    return single_vertex_size;
}


/**
 * Emit a single element using non-DrawArrays protocol.
 */
GLubyte *
emit_element_none( GLubyte * dst,
		   const struct array_state_vector * arrays,
		   unsigned index )
{
    unsigned i;


    for ( i = 0 ; i < arrays->num_arrays ; i++ ) {
	if ( arrays->arrays[i].enabled ) {
	    const size_t offset = index * arrays->arrays[i].true_stride;

	    /* The generic attributes can have more data than is in the
	     * elements.  This is because a vertex array can be a 2 element,
	     * normalized, unsigned short, but the "closest" immediate mode
	     * protocol is for a 4Nus.  Since the sizes are small, the
	     * performance impact on modern processors should be negligible.
	     */
	    (void) memset( dst, 0,
			   ((uint16_t *)arrays->arrays[i].header)[0] );

	    (void) memcpy( dst, arrays->arrays[i].header, 
			   arrays->arrays[i].header_size );

	    dst += arrays->arrays[i].header_size;

	    (void) memcpy( dst, ((GLubyte *) arrays->arrays[i].data) + offset,
			   arrays->arrays[i].element_size );

	    dst += __GLX_PAD( arrays->arrays[i].element_size );
	}
    }

    return dst;
}


/**
 * Emit a single element using "old" DrawArrays protocol from
 * EXT_vertex_arrays / OpenGL 1.1.
 */
GLubyte *
emit_element_old( GLubyte * dst,
		  const struct array_state_vector * arrays,
		  unsigned index )
{
    unsigned i;


    for ( i = 0 ; i < arrays->num_arrays ; i++ ) {
	if ( arrays->arrays[i].enabled ) {
	    const size_t offset = index * arrays->arrays[i].true_stride;

	    (void) memcpy( dst, ((GLubyte *) arrays->arrays[i].data) + offset,
			   arrays->arrays[i].element_size );

	    dst += __GLX_PAD( arrays->arrays[i].element_size );
	}
    }

    return dst;
}


struct array_state *
get_array_entry( const struct array_state_vector * arrays,
		 GLenum key, unsigned index )
{
    unsigned  i;

    for ( i = 0 ; i < arrays->num_arrays ; i++ ) {
	if ( (arrays->arrays[i].key == key)
	     && (arrays->arrays[i].index == index) ) {
	    return & arrays->arrays[i];
	}
    }
    
    return NULL;
}


static GLboolean
allocate_array_info_cache( struct array_state_vector * arrays,
			   size_t required_size )
{
#define MAX_HEADER_SIZE 20
    if ( arrays->array_info_cache_buffer_size < required_size ) {
	GLubyte * temp = realloc( arrays->array_info_cache_base,
				  required_size + MAX_HEADER_SIZE );

	if ( temp == NULL ) {
	    return GL_FALSE;
	}

	arrays->array_info_cache_base = temp;
	arrays->array_info_cache = temp + MAX_HEADER_SIZE;
	arrays->array_info_cache_buffer_size = required_size;
    }

    arrays->array_info_cache_size = required_size;
    return GL_TRUE;
}


/**
 */
void
fill_array_info_cache( struct array_state_vector * arrays )
{
    GLboolean old_DrawArrays_possible;
    unsigned  i;


    /* Determine how many arrays are enabled.
     */

    arrays->enabled_client_array_count = 0;
    old_DrawArrays_possible = arrays->old_DrawArrays_possible;
    for ( i = 0 ; i < arrays->num_arrays ; i++ ) {
	if ( arrays->arrays[i].enabled ) {
	    arrays->enabled_client_array_count++;
	    old_DrawArrays_possible &= arrays->arrays[i].old_DrawArrays_possible;
	}
    }
    
    if ( arrays->new_DrawArrays_possible ) {
	assert( ! arrays->new_DrawArrays_possible );
    }
    else if ( old_DrawArrays_possible ) {
	const size_t required_size = arrays->enabled_client_array_count * 12;
	uint32_t * info;


	if ( ! allocate_array_info_cache( arrays, required_size ) ) {
	    return;
	}


	info = (uint32_t *) arrays->array_info_cache;
	for ( i = 0 ; i < arrays->num_arrays ; i++ ) {
	    if ( arrays->arrays[i].enabled ) {
		*(info++) = arrays->arrays[i].data_type;
		*(info++) = arrays->arrays[i].count;
		*(info++) = arrays->arrays[i].key;
	    }
	}

	arrays->DrawArrays = emit_DrawArrays_old;
	arrays->DrawElements = emit_DrawElements_old;
    }
    else {
	arrays->DrawArrays = emit_DrawArrays_none;
	arrays->DrawElements = emit_DrawElements_none;
    }

    arrays->array_info_cache_valid = GL_TRUE;
}


/**
 * Emit a \c glDrawArrays command using the "none" protocol.  That is,
 * emit immediate-mode commands that are equivalent to the requiested
 * \c glDrawArrays command.  This is used with servers that don't support
 * the OpenGL 1.1 / EXT_vertex_arrays DrawArrays protocol or in cases where
 * vertex state is enabled that is not compatible with that protocol.
 */
void
emit_DrawArrays_none( GLenum mode, GLint first, GLsizei count )
{
    __GLXcontext *gc = __glXGetCurrentContext();
    const __GLXattribute * state = 
       (const __GLXattribute *)(gc->client_state_private);
    struct array_state_vector * arrays = state->array_state;

    size_t single_vertex_size;
    GLubyte * pc;
    unsigned  i;
    static const uint16_t begin_cmd[2] = { 8, X_GLrop_Begin };
    static const uint16_t end_cmd[2]   = { 4, X_GLrop_End };


    single_vertex_size = calculate_single_vertex_size_none( arrays );

    pc = gc->pc;

    (void) memcpy( pc, begin_cmd, 4 );
    *(int *)(pc +  4) = mode;

    pc += 8;

    for ( i = 0 ; i < count ; i++ ) {
	if ( (pc + single_vertex_size) >= gc->bufEnd ) {
	    pc = __glXFlushRenderBuffer(gc, pc);
	}

	pc = emit_element_none( pc, arrays, first + i );
    }

    if ( (pc + 4) >= gc->bufEnd ) {
	pc = __glXFlushRenderBuffer(gc, pc);
    }

    (void) memcpy( pc, end_cmd, 4 );
    pc += 4;

    gc->pc = pc;
    if ( gc->pc > gc->limit ) {
	(void) __glXFlushRenderBuffer(gc, gc->pc);
    }
}


/**
 * Emit the header data for the GL 1.1 / EXT_vertex_arrays DrawArrays
 * protocol.
 * 
 * \param gc                    GLX context.
 * \param arrays                Array state.
 * \param elements_per_request  Location to store the number of elements that
 *                              can fit in a single Render / RenderLarge
 *                              command.
 * \param total_request         Total number of requests for a RenderLarge
 *                              command.  If a Render command is used, this
 *                              will be zero.
 * \param mode                  Drawing mode.
 * \param count                 Number of vertices.
 * 
 * \returns
 * A pointer to the buffer for array data.
 */
static GLubyte *
emit_DrawArrays_header_old( __GLXcontext * gc,
			    struct array_state_vector * arrays,
			    size_t * elements_per_request,
			    unsigned int * total_requests,
			    GLenum mode, GLsizei count )
{
    size_t command_size;
    size_t single_vertex_size;
    const unsigned header_size = 16;
    unsigned  i;
    GLubyte * pc;


    /* Determine the size of the whole command.  This includes the header,
     * the ARRAY_INFO data and the array data.  Once this size is calculated,
     * it will be known whether a Render or RenderLarge command is needed.
     */

    single_vertex_size = 0;
    for ( i = 0 ; i < arrays->num_arrays ; i++ ) {
	if ( arrays->arrays[i].enabled ) {
	    single_vertex_size += __GLX_PAD( arrays->arrays[i].element_size );
	}
    }

    command_size = arrays->array_info_cache_size + header_size 
      + (single_vertex_size * count);


    /* Write the header for either a Render command or a RenderLarge
     * command.  After the header is written, write the ARRAY_INFO data.
     */

    if ( command_size > gc->maxSmallRenderCommandSize ) {
	/* maxSize is the maximum amount of data can be stuffed into a single
	 * packet.  sz_xGLXRenderReq is added because bufSize is the maximum
	 * packet size minus sz_xGLXRenderReq.
	 */
	const size_t maxSize = (gc->bufSize + sz_xGLXRenderReq)
	  - sz_xGLXRenderLargeReq;
	unsigned vertex_requests;


	/* Calculate the number of data packets that will be required to send
	 * the whole command.  To do this, the number of verticies that
	 * will fit in a single buffer must be calculated.
	 * 
	 * The important value here is elements_per_request.  This is the
	 * number of complete array elements that will fit in a single
	 * buffer.  There may be some wasted space at the end of the buffer,
	 * but splitting elements across buffer boundries would be painful.
	 */

	elements_per_request[0] = maxSize / single_vertex_size;

	vertex_requests = (count + elements_per_request[0] - 1)
	  / elements_per_request[0];
	  
	*total_requests = vertex_requests + 1;


	__glXFlushRenderBuffer(gc, gc->pc);

	command_size += 4;

	pc = ((GLubyte *) arrays->array_info_cache) - (header_size + 4);
	*(uint32_t *)(pc +  0) = command_size;
	*(uint32_t *)(pc +  4) = X_GLrop_DrawArrays;
	*(uint32_t *)(pc +  8) = count;
	*(uint32_t *)(pc + 12) = arrays->enabled_client_array_count;
	*(uint32_t *)(pc + 16) = mode;

	__glXSendLargeChunk( gc, 1, *total_requests, pc,
			     header_size + 4 + arrays->array_info_cache_size );

	pc = gc->pc;
    }
    else {
	if ( (gc->pc + command_size) >= gc->bufEnd ) {
	    (void) __glXFlushRenderBuffer(gc, gc->pc);
	}

	pc = gc->pc;
	*(uint16_t *)(pc +  0) = command_size;
	*(uint16_t *)(pc +  2) = X_GLrop_DrawArrays;
	*(uint32_t *)(pc +  4) = count;
	*(uint32_t *)(pc +  8) = arrays->enabled_client_array_count;
	*(uint32_t *)(pc + 12) = mode;

	pc += header_size;

	(void) memcpy( pc, arrays->array_info_cache,
		       arrays->array_info_cache_size );
	pc += arrays->array_info_cache_size;

	*elements_per_request = count;
	*total_requests = 0;
    }


    return pc;
}


/**
 */
void
emit_DrawArrays_old( GLenum mode, GLint first, GLsizei count )
{
    __GLXcontext *gc = __glXGetCurrentContext();
    const __GLXattribute * state = 
       (const __GLXattribute *)(gc->client_state_private);
    struct array_state_vector * arrays = state->array_state;

    GLubyte * pc;
    size_t elements_per_request;
    unsigned total_requests = 0;
    unsigned i;
    size_t total_sent = 0;


    pc = emit_DrawArrays_header_old( gc, arrays, & elements_per_request,
				     & total_requests, mode, count);

    
    /* Write the arrays.
     */

    if ( total_requests == 0 ) {
	assert( elements_per_request >= count );

	for ( i = 0 ; i < count ; i++ ) {
	    pc = emit_element_old( pc, arrays, i + first );
	}

	assert( pc <= gc->bufEnd );

	gc->pc = pc;
	if ( gc->pc > gc->limit ) {
	    (void) __glXFlushRenderBuffer(gc, gc->pc);
	}
    }
    else {
	unsigned req;


	for ( req = 2 ; req <= total_requests ; req++ ) {
	    if ( count < elements_per_request ) {
		elements_per_request = count;
	    }

	    pc = gc->pc;
	    for ( i = 0 ; i < elements_per_request ; i++ ) {
		pc = emit_element_old( pc, arrays, i + first );
	    }

	    first += elements_per_request;

	    total_sent += (size_t) (pc - gc->pc);
	    __glXSendLargeChunk( gc, req, total_requests, gc->pc,
				 pc - gc->pc );

	    count -= elements_per_request;
	}
    }
}


void
emit_DrawElements_none( GLenum mode, GLsizei count, GLenum type,
			const GLvoid *indices )
{
    __GLXcontext *gc = __glXGetCurrentContext();
    const __GLXattribute * state = 
       (const __GLXattribute *)(gc->client_state_private);
    struct array_state_vector * arrays = state->array_state;
    static const uint16_t begin_cmd[2] = { 8, X_GLrop_Begin };
    static const uint16_t end_cmd[2]   = { 4, X_GLrop_End };

    GLubyte * pc;
    size_t single_vertex_size;
    unsigned  i;


    single_vertex_size = calculate_single_vertex_size_none( arrays );


    if ( (gc->pc + single_vertex_size) >= gc->bufEnd ) {
	gc->pc = __glXFlushRenderBuffer(gc, gc->pc);
    }

    pc = gc->pc;

    (void) memcpy( pc, begin_cmd, 4 );
    *(int *)(pc +  4) = mode;

    pc += 8;

    for ( i = 0 ; i < count ; i++ ) {
	unsigned  index = 0;

	if ( (pc + single_vertex_size) >= gc->bufEnd ) {
	    pc = __glXFlushRenderBuffer(gc, pc);
	}

	switch( type ) {
	case GL_UNSIGNED_INT:
	    index = (unsigned) (((GLuint *) indices)[i]);
	    break;
	case GL_UNSIGNED_SHORT:
	    index = (unsigned) (((GLushort *) indices)[i]);
	    break;
	case GL_UNSIGNED_BYTE:
	    index = (unsigned) (((GLubyte *) indices)[i]);
	    break;
	}
	pc = emit_element_none( pc, arrays, index );
    }

    if ( (pc + 4) >= gc->bufEnd ) {
	pc = __glXFlushRenderBuffer(gc, pc);
    }

    (void) memcpy( pc, end_cmd, 4 );
    pc += 4;

    gc->pc = pc;
    if ( gc->pc > gc->limit ) {
	(void) __glXFlushRenderBuffer(gc, gc->pc);
    }
}


/**
 */
void
emit_DrawElements_old( GLenum mode, GLsizei count, GLenum type,
		       const GLvoid *indices )
{
    __GLXcontext *gc = __glXGetCurrentContext();
    const __GLXattribute * state = 
       (const __GLXattribute *)(gc->client_state_private);
    struct array_state_vector * arrays = state->array_state;

    GLubyte * pc;
    size_t elements_per_request;
    unsigned total_requests = 0;
    unsigned i;
    unsigned req;
    unsigned req_element=0;


    pc = emit_DrawArrays_header_old( gc, arrays, & elements_per_request,
				     & total_requests, mode, count);

    
    /* Write the arrays.
     */

    req = 2;
    while ( count > 0 ) {
	if ( count < elements_per_request ) {
	    elements_per_request = count;
	}

	switch( type ) {
	case GL_UNSIGNED_INT: {
	    const GLuint   * ui_ptr = (const GLuint   *) indices + req_element;

	    for ( i = 0 ; i < elements_per_request ; i++ ) {
		const GLint index = (GLint) *(ui_ptr++);
		pc = emit_element_old( pc, arrays, index );
	    }
	    break;
	}
	case GL_UNSIGNED_SHORT: {
	    const GLushort * us_ptr = (const GLushort *) indices + req_element;

	    for ( i = 0 ; i < elements_per_request ; i++ ) {
		const GLint index = (GLint) *(us_ptr++);
		pc = emit_element_old( pc, arrays, index );
	    }
	    break;
	}
	case GL_UNSIGNED_BYTE: {
	    const GLubyte  * ub_ptr = (const GLubyte  *) indices + req_element;

	    for ( i = 0 ; i < elements_per_request ; i++ ) {
		const GLint index = (GLint) *(ub_ptr++);
		pc = emit_element_old( pc, arrays, index );
	    }
	    break;
	}
	}

	if ( total_requests != 0 ) {
	    __glXSendLargeChunk( gc, req, total_requests, gc->pc,
				 pc - gc->pc );
	    pc = gc->pc;
	    req++;
	}

	count -= elements_per_request;
	req_element += elements_per_request;
    }


    assert( (total_requests == 0) || ((req - 1) == total_requests) );

    if ( total_requests == 0 ) {
	assert( pc <= gc->bufEnd );

	gc->pc = pc;
	if ( gc->pc > gc->limit ) {
	    (void) __glXFlushRenderBuffer(gc, gc->pc);
	}
    }
}


/**
 * Validate that the \c mode parameter to \c glDrawArrays, et. al. is valid.
 * If it is not valid, then an error code is set in the GLX context.
 * 
 * \returns
 * \c GL_TRUE if the argument is valid, \c GL_FALSE if is not.
 */
static GLboolean
validate_mode(__GLXcontext *gc, GLenum mode)
{
    switch(mode) {
      case GL_POINTS:
      case GL_LINE_STRIP:
      case GL_LINE_LOOP:
      case GL_LINES:
      case GL_TRIANGLE_STRIP:
      case GL_TRIANGLE_FAN:
      case GL_TRIANGLES:
      case GL_QUAD_STRIP:
      case GL_QUADS:
      case GL_POLYGON:
	break;
      default:
        __glXSetError(gc, GL_INVALID_ENUM);
        return GL_FALSE;
    }
    
    return GL_TRUE;
}


/**
 * Validate that the \c count parameter to \c glDrawArrays, et. al. is valid.
 * A value less than zero is invalid and will result in \c GL_INVALID_VALUE
 * being set.  A value of zero will not result in an error being set, but
 * will result in \c GL_FALSE being returned.
 * 
 * \returns
 * \c GL_TRUE if the argument is valid, \c GL_FALSE if it is not.
 */
static GLboolean
validate_count(__GLXcontext *gc, GLsizei count)
{
    if (count < 0) {
	__glXSetError(gc, GL_INVALID_VALUE);
    }

    return (count > 0);
}


/**
 * Validate that the \c type parameter to \c glDrawElements, et. al. is
 * valid.  Only \c GL_UNSIGNED_BYTE, \c GL_UNSIGNED_SHORT, and
 * \c GL_UNSIGNED_INT are valid.
 *
 * \returns
 * \c GL_TRUE if the argument is valid, \c GL_FALSE if it is not.
 */
static GLboolean validate_type(__GLXcontext *gc, GLenum type)
{
    switch( type ) {
    case GL_UNSIGNED_INT:
    case GL_UNSIGNED_SHORT:
    case GL_UNSIGNED_BYTE:
	return GL_TRUE;
     default:
	__glXSetError(gc, GL_INVALID_ENUM);
	return GL_FALSE;
    }
}


void __indirect_glDrawArrays(GLenum mode, GLint first, GLsizei count)
{
    __GLXcontext *gc = __glXGetCurrentContext();
    const __GLXattribute * state = 
       (const __GLXattribute *)(gc->client_state_private);
    struct array_state_vector * arrays = state->array_state;
    

    if ( validate_mode(gc, mode) && validate_count(gc, count) ) {
	if ( ! arrays->array_info_cache_valid ) {
	    fill_array_info_cache( arrays );
	}

	arrays->DrawArrays(mode, first, count);
    }
}


void __indirect_glArrayElement(GLint index)
{
    __GLXcontext *gc = __glXGetCurrentContext();
    const __GLXattribute * state = 
       (const __GLXattribute *)(gc->client_state_private);
    struct array_state_vector * arrays = state->array_state;

    size_t single_vertex_size;


    single_vertex_size = calculate_single_vertex_size_none( arrays );

    if ( (gc->pc + single_vertex_size) >= gc->bufEnd ) {
	gc->pc = __glXFlushRenderBuffer(gc, gc->pc);
    }

    gc->pc = emit_element_none( gc->pc, arrays, index );

    if ( gc->pc > gc->limit ) {
	(void) __glXFlushRenderBuffer(gc, gc->pc);
    }
}


void __indirect_glDrawElements(GLenum mode, GLsizei count, GLenum type,
		    const GLvoid *indices)
{
    __GLXcontext *gc = __glXGetCurrentContext();
    const __GLXattribute * state = 
       (const __GLXattribute *)(gc->client_state_private);
    struct array_state_vector * arrays = state->array_state;


    if ( validate_mode(gc, mode) && validate_count(gc, count)
	 && validate_type(gc, type) ) {
	if ( ! arrays->array_info_cache_valid ) {
	    fill_array_info_cache( arrays );
	}

	arrays->DrawElements(mode, count, type, indices);
    }
}


void __indirect_glDrawRangeElements(GLenum mode, GLuint start, GLuint end,
				    GLsizei count, GLenum type,
				    const GLvoid *indices)
{
    __GLXcontext *gc = __glXGetCurrentContext();
    const __GLXattribute * state = 
       (const __GLXattribute *)(gc->client_state_private);
    struct array_state_vector * arrays = state->array_state;


    if ( validate_mode(gc, mode) && validate_count(gc, count)
	 && validate_type(gc, type) ) {
	if (end < start) {
	    __glXSetError(gc, GL_INVALID_VALUE);
	    return;
	}

	if ( ! arrays->array_info_cache_valid ) {
	    fill_array_info_cache( arrays );
	}

	arrays->DrawElements(mode, count, type, indices);
    }
}


void __indirect_glMultiDrawArraysEXT(GLenum mode, GLint *first, GLsizei *count,
		       GLsizei primcount)
{
    __GLXcontext *gc = __glXGetCurrentContext();
    const __GLXattribute * state = 
       (const __GLXattribute *)(gc->client_state_private);
    struct array_state_vector * arrays = state->array_state;
    GLsizei  i;


    if ( validate_mode(gc, mode) ) {
	if ( ! arrays->array_info_cache_valid ) {
	    fill_array_info_cache( arrays );
	}

	for ( i = 0 ; i < primcount ; i++ ) {
	    if ( validate_count( gc, count[i] ) ) {
		arrays->DrawArrays(mode, first[i], count[i]);
	    }
	}
    }
}


void __indirect_glMultiDrawElementsEXT(GLenum mode, const GLsizei *count,
			 GLenum type, const GLvoid ** indices,
			 GLsizei primcount)
{
    __GLXcontext *gc = __glXGetCurrentContext();
    const __GLXattribute * state = 
       (const __GLXattribute *)(gc->client_state_private);
    struct array_state_vector * arrays = state->array_state;
    GLsizei  i;


    if ( validate_mode(gc, mode) && validate_type(gc, type) ) {
	if ( ! arrays->array_info_cache_valid ) {
	    fill_array_info_cache( arrays );
	}

	for ( i = 0 ; i < primcount ; i++ ) {
	    if ( validate_count( gc, count[i] ) ) {
		arrays->DrawElements(mode, count[i], type, indices[i]);
	    }
	}
    }
}


#define COMMON_ARRAY_DATA_INIT(a, PTR, TYPE, STRIDE, COUNT, NORMALIZED, HDR_SIZE, OPCODE) \
    do { \
	(a)->data = PTR; \
	(a)->data_type = TYPE; \
	(a)->user_stride = STRIDE; \
	(a)->count = COUNT; \
	(a)->normalized = NORMALIZED; \
	\
	(a)->element_size = __glXTypeSize( TYPE ) * COUNT; \
	(a)->true_stride = (STRIDE == 0) \
	  ? (a)->element_size : STRIDE; \
	\
	(a)->header_size = HDR_SIZE; \
	((uint16_t *) (a)->header)[0] = __GLX_PAD((a)->header_size + (a)->element_size); \
	((uint16_t *) (a)->header)[1] = OPCODE; \
    } while(0)


void __indirect_glVertexPointer( GLint size, GLenum type, GLsizei stride,
				 const GLvoid * pointer )
{
    static const uint16_t short_ops[5] = {
	0, 0, X_GLrop_Vertex2sv, X_GLrop_Vertex3sv, X_GLrop_Vertex4sv
    };
    static const uint16_t int_ops[5] = {
	0, 0, X_GLrop_Vertex2iv, X_GLrop_Vertex3iv, X_GLrop_Vertex4iv
    };
    static const uint16_t float_ops[5] = {
	0, 0, X_GLrop_Vertex2fv, X_GLrop_Vertex3fv, X_GLrop_Vertex4fv
    };
    static const uint16_t double_ops[5] = {
	0, 0, X_GLrop_Vertex2dv, X_GLrop_Vertex3dv, X_GLrop_Vertex4dv
    };
    uint16_t opcode;
    __GLXcontext *gc = __glXGetCurrentContext();
    __GLXattribute * state = (__GLXattribute *)(gc->client_state_private);
    struct array_state_vector * arrays = state->array_state;
    struct array_state * a;


    if (size < 2 || size > 4 || stride < 0) {
        __glXSetError(gc, GL_INVALID_VALUE);
        return;
    }
    
    switch ( type ) {
    case GL_SHORT:	opcode = short_ops[size]; break;
    case GL_INT:	opcode = int_ops[size]; break;
    case GL_FLOAT:	opcode = float_ops[size]; break;
    case GL_DOUBLE:	opcode = double_ops[size]; break;
    default:
        __glXSetError(gc, GL_INVALID_ENUM);
        return;
    }

    a = get_array_entry( arrays, GL_VERTEX_ARRAY, 0 );
    assert( a != NULL );
    COMMON_ARRAY_DATA_INIT( a, pointer, type, stride, size, GL_FALSE, 4,
			    opcode );

    if ( a->enabled ) {
	arrays->array_info_cache_valid = GL_FALSE;
    }
}


void __indirect_glNormalPointer( GLenum type, GLsizei stride,
				 const GLvoid * pointer )
{
    uint16_t opcode;
    __GLXcontext *gc = __glXGetCurrentContext();
    __GLXattribute * state = (__GLXattribute *)(gc->client_state_private);
    struct array_state_vector * arrays = state->array_state;
    struct array_state * a;


    if (stride < 0) {
        __glXSetError(gc, GL_INVALID_VALUE);
        return;
    }

    switch ( type ) {
    case GL_BYTE:	opcode = X_GLrop_Normal3bv; break;
    case GL_SHORT:	opcode = X_GLrop_Normal3sv; break;
    case GL_INT:	opcode = X_GLrop_Normal3iv; break;
    case GL_FLOAT:	opcode = X_GLrop_Normal3fv; break;
    case GL_DOUBLE:	opcode = X_GLrop_Normal3dv; break;
    default:
        __glXSetError(gc, GL_INVALID_ENUM);
        return;
    }

    a = get_array_entry( arrays, GL_NORMAL_ARRAY, 0 );
    assert( a != NULL );
    COMMON_ARRAY_DATA_INIT( a, pointer, type, stride, 3, GL_TRUE, 4,
			    opcode );

    if ( a->enabled ) {
	arrays->array_info_cache_valid = GL_FALSE;
    }
}


void __indirect_glColorPointer( GLint size, GLenum type, GLsizei stride,
				const GLvoid * pointer )
{
    static const uint16_t byte_ops[5] = {
	0, 0, 0, X_GLrop_Color3bv, X_GLrop_Color4bv
    };
    static const uint16_t ubyte_ops[5] = {
	0, 0, 0, X_GLrop_Color3ubv, X_GLrop_Color4ubv
    };
    static const uint16_t short_ops[5] = {
	0, 0, 0, X_GLrop_Color3sv, X_GLrop_Color4sv
    };
    static const uint16_t ushort_ops[5] = {
	0, 0, 0, X_GLrop_Color3usv, X_GLrop_Color4usv
    };
    static const uint16_t int_ops[5] = {
	0, 0, 0, X_GLrop_Color3iv, X_GLrop_Color4iv
    };
    static const uint16_t uint_ops[5] = {
	0, 0, 0, X_GLrop_Color3uiv, X_GLrop_Color4uiv
    };
    static const uint16_t float_ops[5] = {
	0, 0, 0, X_GLrop_Color3fv, X_GLrop_Color4fv
    };
    static const uint16_t double_ops[5] = {
	0, 0, 0, X_GLrop_Color3dv, X_GLrop_Color4dv
    };
    uint16_t opcode;
    __GLXcontext *gc = __glXGetCurrentContext();
    __GLXattribute * state = (__GLXattribute *)(gc->client_state_private);
    struct array_state_vector * arrays = state->array_state;
    struct array_state * a;


    if (size < 3 || size > 4 || stride < 0) {
        __glXSetError(gc, GL_INVALID_VALUE);
        return;
    }
    
    switch ( type ) {
    case GL_BYTE:		opcode = byte_ops[size]; break;
    case GL_UNSIGNED_BYTE:	opcode = ubyte_ops[size]; break;
    case GL_SHORT:		opcode = short_ops[size]; break;
    case GL_UNSIGNED_SHORT:	opcode = ushort_ops[size]; break;
    case GL_INT:		opcode = int_ops[size]; break;
    case GL_UNSIGNED_INT:	opcode = uint_ops[size]; break;
    case GL_FLOAT:		opcode = float_ops[size]; break;
    case GL_DOUBLE:		opcode = double_ops[size]; break;
    default:
        __glXSetError(gc, GL_INVALID_ENUM);
        return;
    }

    a = get_array_entry( arrays, GL_COLOR_ARRAY, 0 );
    assert( a != NULL );
    COMMON_ARRAY_DATA_INIT( a, pointer, type, stride, size, GL_TRUE, 4,
			    opcode );

    if ( a->enabled ) {
	arrays->array_info_cache_valid = GL_FALSE;
    }
}


void __indirect_glIndexPointer( GLenum type, GLsizei stride,
				const GLvoid * pointer )
{
    uint16_t opcode;
    __GLXcontext *gc = __glXGetCurrentContext();
    __GLXattribute * state = (__GLXattribute *)(gc->client_state_private);
    struct array_state_vector * arrays = state->array_state;
    struct array_state * a;


    if (stride < 0) {
        __glXSetError(gc, GL_INVALID_VALUE);
        return;
    }
    
    switch ( type ) {
    case GL_UNSIGNED_BYTE:	opcode = X_GLrop_Indexubv; break;
    case GL_SHORT:		opcode = X_GLrop_Indexsv; break;
    case GL_INT:		opcode = X_GLrop_Indexiv; break;
    case GL_FLOAT:		opcode = X_GLrop_Indexfv; break;
    case GL_DOUBLE:		opcode = X_GLrop_Indexdv; break;
    default:
        __glXSetError(gc, GL_INVALID_ENUM);
        return;
    }

    a = get_array_entry( arrays, GL_INDEX_ARRAY, 0 );
    assert( a != NULL );
    COMMON_ARRAY_DATA_INIT( a, pointer, type, stride, 1, GL_FALSE, 4,
			    opcode );

    if ( a->enabled ) {
	arrays->array_info_cache_valid = GL_FALSE;
    }
}


void __indirect_glEdgeFlagPointer( GLsizei stride, const GLvoid * pointer )
{
    __GLXcontext *gc = __glXGetCurrentContext();
    __GLXattribute * state = (__GLXattribute *)(gc->client_state_private);
    struct array_state_vector * arrays = state->array_state;
    struct array_state * a;


    if (stride < 0) {
        __glXSetError(gc, GL_INVALID_VALUE);
        return;
    }
    

    a = get_array_entry( arrays, GL_EDGE_FLAG_ARRAY, 0 );
    assert( a != NULL );
    COMMON_ARRAY_DATA_INIT( a, pointer, GL_UNSIGNED_BYTE, stride, 1, GL_FALSE,
			    4, X_GLrop_EdgeFlagv );

    if ( a->enabled ) {
	arrays->array_info_cache_valid = GL_FALSE;
    }
}


void __indirect_glTexCoordPointer( GLint size, GLenum type, GLsizei stride,
				   const GLvoid * pointer )
{
    static const uint16_t short_ops[5] = {
	0, X_GLrop_TexCoord1sv, X_GLrop_TexCoord2sv, X_GLrop_TexCoord3sv, X_GLrop_TexCoord4sv
    };
    static const uint16_t int_ops[5] = {
	0, X_GLrop_TexCoord1iv, X_GLrop_TexCoord2iv, X_GLrop_TexCoord3iv, X_GLrop_TexCoord4iv
    };
    static const uint16_t float_ops[5] = {
	0, X_GLrop_TexCoord1dv, X_GLrop_TexCoord2fv, X_GLrop_TexCoord3fv, X_GLrop_TexCoord4fv
    };
    static const uint16_t double_ops[5] = {
	0, X_GLrop_TexCoord1dv, X_GLrop_TexCoord2dv, X_GLrop_TexCoord3dv, X_GLrop_TexCoord4dv
    };

    static const uint16_t mshort_ops[5] = {
	0, X_GLrop_MultiTexCoord1svARB, X_GLrop_MultiTexCoord2svARB, X_GLrop_MultiTexCoord3svARB, X_GLrop_MultiTexCoord4svARB
    };
    static const uint16_t mint_ops[5] = {
	0, X_GLrop_MultiTexCoord1ivARB, X_GLrop_MultiTexCoord2ivARB, X_GLrop_MultiTexCoord3ivARB, X_GLrop_MultiTexCoord4ivARB
    };
    static const uint16_t mfloat_ops[5] = {
	0, X_GLrop_MultiTexCoord1dvARB, X_GLrop_MultiTexCoord2fvARB, X_GLrop_MultiTexCoord3fvARB, X_GLrop_MultiTexCoord4fvARB
    };
    static const uint16_t mdouble_ops[5] = {
	0, X_GLrop_MultiTexCoord1dvARB, X_GLrop_MultiTexCoord2dvARB, X_GLrop_MultiTexCoord3dvARB, X_GLrop_MultiTexCoord4dvARB
    };

    uint16_t opcode;
    __GLXcontext *gc = __glXGetCurrentContext();
    __GLXattribute * state = (__GLXattribute *)(gc->client_state_private);
    struct array_state_vector * arrays = state->array_state;
    struct array_state * a;
    unsigned header_size;
    unsigned index;


    if (size < 1 || size > 4 || stride < 0) {
        __glXSetError(gc, GL_INVALID_VALUE);
        return;
    }
    
    index = arrays->active_texture_unit;
    if ( index == 0 ) {
	switch ( type ) {
	case GL_SHORT:		opcode = short_ops[size]; break;
	case GL_INT:		opcode = int_ops[size]; break;
	case GL_FLOAT:		opcode = float_ops[size]; break;
	case GL_DOUBLE:		opcode = double_ops[size]; break;
	default:
	    __glXSetError(gc, GL_INVALID_ENUM);
	    return;
	}

	header_size = 4;
    }
    else {
	switch ( type ) {
	case GL_SHORT:		opcode = mshort_ops[size]; break;
	case GL_INT:		opcode = mint_ops[size]; break;
	case GL_FLOAT:		opcode = mfloat_ops[size]; break;
	case GL_DOUBLE:		opcode = mdouble_ops[size]; break;
	default:
	    __glXSetError(gc, GL_INVALID_ENUM);
	    return;
	}

	header_size = 8;
    }

    a = get_array_entry( arrays, GL_TEXTURE_COORD_ARRAY, index );
    assert( a != NULL );
    COMMON_ARRAY_DATA_INIT( a, pointer, type, stride, size, GL_FALSE,
			    header_size, opcode );

    if ( a->enabled ) {
	arrays->array_info_cache_valid = GL_FALSE;
    }
}


void __indirect_glSecondaryColorPointerEXT( GLint size, GLenum type, GLsizei stride,
				const GLvoid * pointer )
{
    uint16_t opcode;
    __GLXcontext *gc = __glXGetCurrentContext();
    __GLXattribute * state = (__GLXattribute *)(gc->client_state_private);
    struct array_state_vector * arrays = state->array_state;
    struct array_state * a;


    if (size != 3 || stride < 0) {
        __glXSetError(gc, GL_INVALID_VALUE);
        return;
    }
    
    switch ( type ) {
    case GL_BYTE:		opcode = 4126; break;
    case GL_UNSIGNED_BYTE:	opcode = 4131; break;
    case GL_SHORT:		opcode = 4127; break;
    case GL_UNSIGNED_SHORT:	opcode = 4132; break;
    case GL_INT:		opcode = 4128; break;
    case GL_UNSIGNED_INT:	opcode = 4133; break;
    case GL_FLOAT:		opcode = 4129; break;
    case GL_DOUBLE:		opcode = 4130; break;
    default:
        __glXSetError(gc, GL_INVALID_ENUM);
        return;
    }

    a = get_array_entry( arrays, GL_SECONDARY_COLOR_ARRAY, 0 );
    if ( a == NULL ) {
        __glXSetError(gc, GL_INVALID_OPERATION);
        return;
    }

    COMMON_ARRAY_DATA_INIT( a, pointer, type, stride, size, GL_TRUE, 4,
			    opcode );

    if ( a->enabled ) {
	arrays->array_info_cache_valid = GL_FALSE;
    }
}


void __indirect_glFogCoordPointerEXT( GLenum type, GLsizei stride,
				const GLvoid * pointer )
{
    uint16_t opcode;
    __GLXcontext *gc = __glXGetCurrentContext();
    __GLXattribute * state = (__GLXattribute *)(gc->client_state_private);
    struct array_state_vector * arrays = state->array_state;
    struct array_state * a;


    if (stride < 0) {
        __glXSetError(gc, GL_INVALID_VALUE);
        return;
    }

    switch ( type ) {
    case GL_FLOAT:		opcode = 4124; break;
    case GL_DOUBLE:		opcode = 4125; break;
    default:
        __glXSetError(gc, GL_INVALID_ENUM);
        return;
    }

    a = get_array_entry( arrays, GL_FOG_COORD_ARRAY, 0 );
    if ( a == NULL ) {
        __glXSetError(gc, GL_INVALID_OPERATION);
        return;
    }

    COMMON_ARRAY_DATA_INIT( a, pointer, type, stride, 1, GL_FALSE, 4,
			    opcode );

    if ( a->enabled ) {
	arrays->array_info_cache_valid = GL_FALSE;
    }
}


void __indirect_glVertexAttribPointerARB(GLuint index, GLint size,
					 GLenum type, GLboolean normalized,
					 GLsizei stride,
					 const GLvoid * pointer)
{
    static const uint16_t short_ops[5]  = { 0, 4189, 4190, 4191, 4192 };
    static const uint16_t float_ops[5]  = { 0, 4193, 4194, 4195, 4196 };
    static const uint16_t double_ops[5] = { 0, 4197, 4198, 4199, 4200 };

    uint16_t opcode;
    __GLXcontext *gc = __glXGetCurrentContext();
    __GLXattribute * state = (__GLXattribute *)(gc->client_state_private);
    struct array_state_vector * arrays = state->array_state;
    struct array_state * a;
    unsigned true_immediate_count;
    unsigned true_immediate_size;


    if ( (size < 1) || (size > 4) || (stride < 0) 
	 || (index > arrays->num_vertex_program_attribs) ){
        __glXSetError(gc, GL_INVALID_VALUE);
        return;
    }

    if ( normalized && (type != GL_FLOAT) && (type != GL_DOUBLE)) {
	switch( type ) {
	case GL_BYTE:           opcode = X_GLrop_VertexAttrib4NbvARB;  break;
	case GL_UNSIGNED_BYTE:  opcode = X_GLrop_VertexAttrib4NubvARB; break;
	case GL_SHORT:          opcode = X_GLrop_VertexAttrib4NsvARB;  break;
	case GL_UNSIGNED_SHORT: opcode = X_GLrop_VertexAttrib4NusvARB; break;
	case GL_INT:            opcode = X_GLrop_VertexAttrib4NivARB;  break;
	case GL_UNSIGNED_INT:   opcode = X_GLrop_VertexAttrib4NuivARB; break;
	default:
	    __glXSetError(gc, GL_INVALID_ENUM);
	    return;
	}
	
	true_immediate_count = 4;
    }
    else {
	true_immediate_count = size;

	switch( type ) {
	case GL_BYTE:
	    opcode = X_GLrop_VertexAttrib4bvARB;
	    true_immediate_count = 4;
	    break;
	case GL_UNSIGNED_BYTE:
	    opcode = X_GLrop_VertexAttrib4ubvARB;
	    true_immediate_count = 4;
	    break;
	case GL_SHORT:
	    opcode = short_ops[size];
	    break;
	case GL_UNSIGNED_SHORT:
	    opcode = X_GLrop_VertexAttrib4usvARB;
	    true_immediate_count = 4;
	    break;
	case GL_INT:
            opcode = X_GLrop_VertexAttrib4ivARB;
	    true_immediate_count = 4;
	    break;
	case GL_UNSIGNED_INT:
	    opcode = X_GLrop_VertexAttrib4uivARB;
	    true_immediate_count = 4;
	    break;
	case GL_FLOAT:
	    opcode = float_ops[size];
	    break;
	case GL_DOUBLE:
	    opcode = double_ops[size];
	    break;
	default:
	    __glXSetError(gc, GL_INVALID_ENUM);
	    return;
	}
    }

    a = get_array_entry( arrays, GL_VERTEX_ATTRIB_ARRAY_POINTER, index );
    if ( a == NULL ) {
        __glXSetError(gc, GL_INVALID_OPERATION);
        return;
    }

    COMMON_ARRAY_DATA_INIT( a, pointer, type, stride, size, normalized, 8,
			    opcode );

    true_immediate_size = __glXTypeSize(type) * true_immediate_count;
    ((uint16_t *) (a)->header)[0] = __GLX_PAD(a->header_size 
					      + true_immediate_size);

    if ( a->enabled ) {
	arrays->array_info_cache_valid = GL_FALSE;
    }
}


/**
 * I don't have 100% confidence that this is correct.  The different rules
 * about whether or not generic vertex attributes alias "classic" vertex
 * attributes (i.e., attrib1 ?= primary color) between ARB_vertex_program,
 * ARB_vertex_shader, and NV_vertex_program are a bit confusing.  My
 * feeling is that the client-side doesn't have to worry about it.  The
 * client just sends all the data to the server and lets the server deal
 * with it.
 */
void __indirect_glVertexAttribPointerNV( GLuint index, GLint size,
					 GLenum type, GLsizei stride,
					 const GLvoid * pointer)
{
    __GLXcontext *gc = __glXGetCurrentContext();
    GLboolean normalized = GL_FALSE;


    switch( type ) {
    case GL_UNSIGNED_BYTE:
	if ( size != 4 ) {
	    __glXSetError(gc, GL_INVALID_VALUE);
	    return;
	}
	normalized = GL_TRUE;

    case GL_SHORT:
    case GL_FLOAT:
    case GL_DOUBLE:
	__indirect_glVertexAttribPointerARB(index, size, type,
					    normalized,
					    stride, pointer);
	return;
    default:
	__glXSetError(gc, GL_INVALID_ENUM);
	return;
    }
}


void __indirect_glClientActiveTextureARB(GLenum texture)
{
    __GLXcontext * const gc = __glXGetCurrentContext();
    __GLXattribute * const state = (__GLXattribute *)(gc->client_state_private);
    struct array_state_vector * const arrays = state->array_state;
    const GLint unit = (GLint) texture - GL_TEXTURE0;


    if ( (unit < 0) || (unit >= arrays->num_texture_units) ) {
	__glXSetError(gc, GL_INVALID_ENUM);
	return;
    }

    arrays->active_texture_unit = unit;
}


/**
 * Modify the enable state for the selected array
 */
GLboolean
__glXSetArrayEnable(__GLXattribute *state, GLenum key, unsigned index,
                    GLboolean enable)
{
    struct array_state_vector * arrays = state->array_state;
    struct array_state * a;
    

    /* Texture coordinate arrays have an implict index set when the
     * application calls glClientActiveTexture.
     */
    if (key == GL_TEXTURE_COORD_ARRAY) {
        index = arrays->active_texture_unit;
    }

    a = get_array_entry( arrays, key, index );

    if ( (a != NULL) && (a->enabled != enable) ) {
	a->enabled = enable;
	arrays->array_info_cache_valid = GL_FALSE;
    }

    return (a != NULL);
}


void
__glXArrayDisableAll( __GLXattribute * state )
{
    struct array_state_vector * arrays = state->array_state;
    unsigned  i;


    for ( i = 0 ; i < arrays->num_arrays ; i++ ) {
	arrays->arrays[i].enabled = GL_FALSE;
    }

    arrays->array_info_cache_valid = GL_FALSE;
}


/**
 */
GLboolean
__glXGetArrayEnable( const __GLXattribute * const state,
		     GLenum key, unsigned index, GLintptr * dest )
{
    const struct array_state_vector * arrays = state->array_state;
    const struct array_state * a = get_array_entry( (struct array_state_vector *) arrays,
						    key, index );

    if ( a != NULL ) {
	*dest = (GLintptr) a->enabled;
    }

    return (a != NULL);
}


/**
 */
GLboolean
__glXGetArrayType( const __GLXattribute * const state,
		     GLenum key, unsigned index, GLintptr * dest )
{
    const struct array_state_vector * arrays = state->array_state;
    const struct array_state * a = get_array_entry( (struct array_state_vector *) arrays,
						    key, index );

    if ( a != NULL ) {
	*dest = (GLintptr) a->data_type;
    }

    return (a != NULL);
}


/**
 */
GLboolean
__glXGetArraySize( const __GLXattribute * const state,
		     GLenum key, unsigned index, GLintptr * dest )
{
    const struct array_state_vector * arrays = state->array_state;
    const struct array_state * a = get_array_entry( (struct array_state_vector *) arrays,
						    key, index );

    if ( a != NULL ) {
	*dest = (GLintptr) a->count;
    }

    return (a != NULL);
}


/**
 */
GLboolean
__glXGetArrayStride( const __GLXattribute * const state,
		     GLenum key, unsigned index, GLintptr * dest )
{
    const struct array_state_vector * arrays = state->array_state;
    const struct array_state * a = get_array_entry( (struct array_state_vector *) arrays,
						    key, index );

    if ( a != NULL ) {
	*dest = (GLintptr) a->user_stride;
    }

    return (a != NULL);
}


/**
 */
GLboolean
__glXGetArrayPointer( const __GLXattribute * const state,
		      GLenum key, unsigned index, void ** dest )
{
    const struct array_state_vector * arrays = state->array_state;
    const struct array_state * a = get_array_entry( (struct array_state_vector *) arrays,
						    key, index );


    if ( a != NULL ) {
	*dest = (void *) (a->data);
    }

    return (a != NULL);
}


/**
 */
GLboolean
__glXGetArrayNormalized( const __GLXattribute * const state,
			 GLenum key, unsigned index, GLintptr * dest )
{
    const struct array_state_vector * arrays = state->array_state;
    const struct array_state * a = get_array_entry( (struct array_state_vector *) arrays,
						    key, index );


    if ( a != NULL ) {
	*dest = (GLintptr) a->normalized;
    }

    return (a != NULL);
}


/**
 */
GLuint
__glXGetActiveTextureUnit( const __GLXattribute * const state )
{
    return state->array_state->active_texture_unit;
}


void
__glXPushArrayState( __GLXattribute * state )
{
    struct array_state_vector * arrays = state->array_state;
    struct array_stack_state * stack = & arrays->stack[ (arrays->stack_index * arrays->num_arrays)];
    unsigned  i;

    /* XXX are we pushing _all_ the necessary fields? */
    for ( i = 0 ; i < arrays->num_arrays ; i++ ) {
	stack[i].data = arrays->arrays[i].data;
	stack[i].data_type = arrays->arrays[i].data_type;
	stack[i].user_stride = arrays->arrays[i].user_stride;
	stack[i].count = arrays->arrays[i].count;
	stack[i].key = arrays->arrays[i].key;
        stack[i].index = arrays->arrays[i].index;
	stack[i].enabled = arrays->arrays[i].enabled;
    }

    arrays->active_texture_unit_stack[ arrays->stack_index ] =
      arrays->active_texture_unit;

    arrays->stack_index++;
}


void
__glXPopArrayState( __GLXattribute * state )
{
    struct array_state_vector * arrays = state->array_state;
    struct array_stack_state * stack;
    unsigned  i;


    arrays->stack_index--;
    stack = & arrays->stack[ (arrays->stack_index * arrays->num_arrays) ];

    for ( i = 0 ; i < arrays->num_arrays ; i++ ) {
	switch ( stack[i].key ) {
	case GL_NORMAL_ARRAY:
	    __indirect_glNormalPointer( stack[i].data_type,
					stack[i].user_stride,
					stack[i].data );
	    break;
	case GL_COLOR_ARRAY:
	    __indirect_glColorPointer( stack[i].count,
				       stack[i].data_type,
				       stack[i].user_stride,
				       stack[i].data );
	    break;
	case GL_INDEX_ARRAY:
	    __indirect_glIndexPointer( stack[i].data_type,
				       stack[i].user_stride,
				       stack[i].data );
	    break;
	case GL_EDGE_FLAG_ARRAY:
	    __indirect_glEdgeFlagPointer( stack[i].user_stride,
					  stack[i].data );
	    break;
	case GL_TEXTURE_COORD_ARRAY:
	    arrays->active_texture_unit = stack[i].index;
	    __indirect_glTexCoordPointer( stack[i].count,
					  stack[i].data_type,
					  stack[i].user_stride,
					  stack[i].data );
	    break;
	case GL_SECONDARY_COLOR_ARRAY:
	    __indirect_glSecondaryColorPointerEXT( stack[i].count,
						   stack[i].data_type,
						   stack[i].user_stride,
						   stack[i].data );
	    break;
	case GL_FOG_COORDINATE_ARRAY:
	    __indirect_glFogCoordPointerEXT( stack[i].data_type,
					     stack[i].user_stride,
					     stack[i].data );
	    break;

	}
	
	__glXSetArrayEnable( state, stack[i].key, stack[i].index,
			     stack[i].enabled );
    }

    arrays->active_texture_unit = 
      arrays->active_texture_unit_stack[ arrays->stack_index ];
}