glxcmds.c   [plain text]

/*
 * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
 * Copyright (C) 1991-2000 Silicon Graphics, 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, 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 including the dates of first publication and
 * either this permission notice or a reference to
 * http://oss.sgi.com/projects/FreeB/
 * 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
 * SILICON GRAPHICS, INC. 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.
 *
 * Except as contained in this notice, the name of Silicon Graphics, Inc.
 * shall not be used in advertising or otherwise to promote the sale, use or
 * other dealings in this Software without prior written authorization from
 * Silicon Graphics, Inc.
 */

#ifdef HAVE_DIX_CONFIG_H
#include <dix-config.h>
#endif

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

#include "glxserver.h"
#include <GL/glxtokens.h>
#include <unpack.h>
#include <pixmapstr.h>
#include <windowstr.h>
#include "glxutil.h"
#include "glxext.h"
#include "glapitable.h"
#include "glapi.h"
#include "glthread.h"
#include "dispatch.h"
#include "indirect_dispatch.h"
#include "indirect_table.h"
#include "indirect_util.h"

static int
validGlxScreen(ClientPtr client, int screen, __GLXscreen **pGlxScreen, int *err)
{
    /*
    ** Check if screen exists.
    */
    if (screen < 0 || screen >= screenInfo.numScreens) {
	client->errorValue = screen;
	*err = BadValue;
	return FALSE;
    }
    *pGlxScreen = glxGetScreen(screenInfo.screens[screen]);

    return TRUE;
}

static int
validGlxFBConfig(ClientPtr client, __GLXscreen *pGlxScreen, XID id,
		 __GLXconfig **config, int *err)
{
    __GLXconfig *m;

    for (m = pGlxScreen->fbconfigs; m != NULL; m = m->next)
	if (m->fbconfigID == id) {
	    *config = m;
	    return TRUE;
	}

    client->errorValue = id;
    *err = __glXError(GLXBadFBConfig);

    return FALSE;
}

static int
validGlxVisual(ClientPtr client, __GLXscreen *pGlxScreen, XID id,
	       __GLXconfig **config, int *err)
{
    int i;

    for (i = 0; i < pGlxScreen->numVisuals; i++)
 	if (pGlxScreen->visuals[i]->visualID == id) {
	    *config = pGlxScreen->visuals[i];
	    return TRUE;
	}

    client->errorValue = id;
    *err = BadValue;

    return FALSE;
}

static int
validGlxFBConfigForWindow(ClientPtr client, __GLXconfig *config,
			  DrawablePtr pDraw, int *err)
{
    ScreenPtr pScreen = pDraw->pScreen;
    VisualPtr pVisual = NULL;
    XID vid;
    int i;

    vid = wVisual((WindowPtr)pDraw);
    for (i = 0; i < pScreen->numVisuals; i++) {
	if (pScreen->visuals[i].vid == vid) {
	    pVisual = &pScreen->visuals[i];
	    break;
	}
    }

    /* FIXME: What exactly should we check here... */
    if (pVisual->class != glxConvertToXVisualType(config->visualType) ||
	!(config->drawableType & GLX_WINDOW_BIT)) {
	client->errorValue = pDraw->id;
	*err = BadMatch;
	return FALSE;
    }

    return TRUE;
}

static int
validGlxContext(ClientPtr client, XID id, int access_mode,
		__GLXcontext **context, int *err)
{
    *err = dixLookupResourceByType((pointer *) context, id,
				   __glXContextRes, client, access_mode);
    if (*err != Success || (*context)->idExists == GL_FALSE) {
	client->errorValue = id;
	if (*err == BadValue || *err == Success)
	    *err = __glXError(GLXBadContext);
	return FALSE;
    }

    return TRUE;
}

static int
validGlxDrawable(ClientPtr client, XID id, int type, int access_mode,
		 __GLXdrawable **drawable, int *err)
{
    int rc;

    rc = dixLookupResourceByType((pointer *) drawable, id,
				 __glXDrawableRes, client, access_mode);
    if (rc != Success && rc != BadValue) {
	*err = rc;
	client->errorValue = id;
	return FALSE;
    }

    /* If the ID of the glx drawable we looked up doesn't match the id
     * we looked for, it's because we looked it up under the X
     * drawable ID (see DoCreateGLXDrawable). */
    if (rc == BadValue ||
	(*drawable)->drawId != id ||
	(type != GLX_DRAWABLE_ANY && type != (*drawable)->type)) {
	client->errorValue = id;
	switch (type) {
	case GLX_DRAWABLE_WINDOW:
	    *err = __glXError(GLXBadWindow);
	    return FALSE;
	case GLX_DRAWABLE_PIXMAP:
	    *err = __glXError(GLXBadPixmap);
	    return FALSE;
	case GLX_DRAWABLE_PBUFFER:
	    *err = __glXError(GLXBadPbuffer);
	    return FALSE;
	case GLX_DRAWABLE_ANY:
	    *err = __glXError(GLXBadDrawable);
	    return FALSE;
	}
    }

    return TRUE;
}

void
__glXContextDestroy(__GLXcontext *context)
{
    __glXFlushContextCache();
}

static void __glXdirectContextDestroy(__GLXcontext *context)
{
    __glXContextDestroy(context);
    free(context);
}

static __GLXcontext *__glXdirectContextCreate(__GLXscreen *screen,
					      __GLXconfig *modes,
					      __GLXcontext *shareContext)
{
    __GLXcontext *context;

    context = calloc(1, sizeof (__GLXcontext));
    if (context == NULL)
	return NULL;

    context->destroy = __glXdirectContextDestroy;

    return context;
}

/**
 * Create a GL context with the given properties.  This routine is used
 * to implement \c glXCreateContext, \c glXCreateNewContext, and
 * \c glXCreateContextWithConfigSGIX.  This works becuase of the hack way
 * that GLXFBConfigs are implemented.  Basically, the FBConfigID is the
 * same as the VisualID.
 */

static int
DoCreateContext(__GLXclientState *cl, GLXContextID gcId,
		GLXContextID shareList, __GLXconfig *config,
		__GLXscreen *pGlxScreen, GLboolean isDirect)
{
    ClientPtr client = cl->client;
    __GLXcontext *glxc, *shareglxc;
    int err;
    
    LEGAL_NEW_RESOURCE(gcId, client);

    /*
    ** Find the display list space that we want to share.  
    **
    ** NOTE: In a multithreaded X server, we would need to keep a reference
    ** count for each display list so that if one client detroyed a list that 
    ** another client was using, the list would not really be freed until it 
    ** was no longer in use.  Since this sample implementation has no support 
    ** for multithreaded servers, we don't do this.  
    */
    if (shareList == None) {
	shareglxc = 0;
    } else {
	if (!validGlxContext(client, shareList, DixReadAccess,
			     &shareglxc, &err))
	    return err;

	if (shareglxc->isDirect) {
	    /*
	    ** NOTE: no support for sharing display lists between direct
	    ** contexts, even if they are in the same address space.
	    */
#if 0
            /* Disabling this code seems to allow shared display lists
             * and texture objects to work.  We'll leave it disabled for now.
             */
	    client->errorValue = shareList;
	    return BadMatch;
#endif
	} else {
	    /*
	    ** Create an indirect context regardless of what the client asked
	    ** for; this way we can share display list space with shareList.
	    */
	    isDirect = GL_FALSE;
	}
    }

    /*
    ** Allocate memory for the new context
    */
    if (!isDirect)
	glxc = pGlxScreen->createContext(pGlxScreen, config, shareglxc);
    else
	glxc = __glXdirectContextCreate(pGlxScreen, config, shareglxc);
    if (!glxc) {
	return BadAlloc;
    }

    /*
    ** Initially, setup the part of the context that could be used by
    ** a GL core that needs windowing information (e.g., Mesa).
    */
    glxc->pGlxScreen = pGlxScreen;
    glxc->config = config;

    /*
    ** Register this context as a resource.
    */
    if (!AddResource(gcId, __glXContextRes, (pointer)glxc)) {
	(*glxc->destroy)(glxc);
	client->errorValue = gcId;
	return BadAlloc;
    }
    
    /*
    ** Finally, now that everything is working, setup the rest of the
    ** context.
    */
    glxc->id = gcId;
    glxc->share_id = shareList;
    glxc->idExists = GL_TRUE;
    glxc->isCurrent = GL_FALSE;
    glxc->isDirect = isDirect;
    glxc->renderMode = GL_RENDER;

    __glXAddToContextList(glxc);

    return Success;
}

int __glXDisp_CreateContext(__GLXclientState *cl, GLbyte *pc)
{
    ClientPtr client = cl->client;
    xGLXCreateContextReq *req = (xGLXCreateContextReq *) pc;
    __GLXconfig *config;
    __GLXscreen *pGlxScreen;
    int err;

    REQUEST_SIZE_MATCH(xGLXCreateContextReq);

    if (!validGlxScreen(cl->client, req->screen, &pGlxScreen, &err))
	return err;
    if (!validGlxVisual(cl->client, pGlxScreen, req->visual, &config, &err))
	return err;

    return DoCreateContext(cl, req->context, req->shareList,
			   config, pGlxScreen, req->isDirect);
}

int __glXDisp_CreateNewContext(__GLXclientState *cl, GLbyte *pc)
{
    ClientPtr client = cl->client;
    xGLXCreateNewContextReq *req = (xGLXCreateNewContextReq *) pc;
    __GLXconfig *config;
    __GLXscreen *pGlxScreen;
    int err;

    REQUEST_SIZE_MATCH(xGLXCreateNewContextReq);

    if (!validGlxScreen(cl->client, req->screen, &pGlxScreen, &err))
	return err;
    if (!validGlxFBConfig(cl->client, pGlxScreen, req->fbconfig, &config, &err))
	return err;

    return DoCreateContext(cl, req->context, req->shareList,
			   config, pGlxScreen, req->isDirect);
}

int __glXDisp_CreateContextWithConfigSGIX(__GLXclientState *cl, GLbyte *pc)
{
    ClientPtr client = cl->client;
    xGLXCreateContextWithConfigSGIXReq *req = 
	(xGLXCreateContextWithConfigSGIXReq *) pc;
    __GLXconfig *config;
    __GLXscreen *pGlxScreen;
    int err;

    REQUEST_SIZE_MATCH(xGLXCreateContextWithConfigSGIXReq);

    if (!validGlxScreen(cl->client, req->screen, &pGlxScreen, &err))
	return err;
    if (!validGlxFBConfig(cl->client, pGlxScreen, req->fbconfig, &config, &err))
	return err;

    return DoCreateContext(cl, req->context, req->shareList,
			   config, pGlxScreen, req->isDirect);
}

int __glXDisp_DestroyContext(__GLXclientState *cl, GLbyte *pc)
{
    ClientPtr client = cl->client;
    xGLXDestroyContextReq *req = (xGLXDestroyContextReq *) pc;
    __GLXcontext *glxc;
    int err;

    REQUEST_SIZE_MATCH(xGLXDestroyContextReq);

    if (!validGlxContext(cl->client, req->context, DixDestroyAccess,
			 &glxc, &err))
	    return err;

    glxc->idExists = GL_FALSE;
    if (!glxc->isCurrent)
        FreeResourceByType(req->context, __glXContextRes, FALSE);

    return Success;
}

/*
 * This will return "deleted" contexts, ie, where idExists is GL_FALSE.
 * Contrast validGlxContext, which will not.  We're cheating here and
 * using the XID as the context tag, which is fine as long as we defer
 * actually destroying the context until it's no longer referenced, and
 * block clients from trying to MakeCurrent on contexts that are on the
 * way to destruction.  Notice that DoMakeCurrent calls validGlxContext
 * for new contexts but __glXLookupContextByTag for previous contexts.
 */
__GLXcontext *__glXLookupContextByTag(__GLXclientState *cl, GLXContextTag tag)
{
    __GLXcontext *ret;

    if (dixLookupResourceByType((void **)&ret, tag, __glXContextRes,
                                cl->client, DixUseAccess) == Success)
        return ret;

    return NULL;
}

/*****************************************************************************/

static void StopUsingContext(__GLXcontext *glxc)
{
    if (glxc) {
	if (glxc == __glXLastContext) {
	    /* Tell server GL library */
	    __glXLastContext = 0;
	}
	glxc->isCurrent = GL_FALSE;
	if (!glxc->idExists) {
            FreeResourceByType(glxc->id, __glXContextRes, FALSE);
	}
    }
}

static void StartUsingContext(__GLXclientState *cl, __GLXcontext *glxc)
{
    glxc->isCurrent = GL_TRUE;
    __glXLastContext = glxc;	
}

/**
 * This is a helper function to handle the legacy (pre GLX 1.3) cases
 * where passing an X window to glXMakeCurrent is valid.  Given a
 * resource ID, look up the GLX drawable if available, otherwise, make
 * sure it's an X window and create a GLX drawable one the fly.
 */
static __GLXdrawable *
__glXGetDrawable(__GLXcontext *glxc, GLXDrawable drawId, ClientPtr client,
		 int *error)
{
    DrawablePtr pDraw;
    __GLXdrawable *pGlxDraw;
    int rc;

    if (validGlxDrawable(client, drawId, GLX_DRAWABLE_ANY,
			 DixWriteAccess, &pGlxDraw, &rc)) {
	if (glxc != NULL && pGlxDraw->config != glxc->config) {
	    client->errorValue = drawId;
	    *error = BadMatch;
	    return NULL;
	}

	return pGlxDraw;
    }

    /* No active context and an unknown drawable, bail. */
    if (glxc == NULL) {
	    client->errorValue = drawId;
	    *error = BadMatch;
	    return NULL;
    }

    /* The drawId wasn't a GLX drawable.  Make sure it's a window and
     * create a GLXWindow for it.  Check that the drawable screen
     * matches the context screen and that the context fbconfig is
     * compatible with the window visual. */

    rc = dixLookupDrawable(&pDraw, drawId, client, 0, DixGetAttrAccess);
    if (rc != Success || pDraw->type != DRAWABLE_WINDOW) {
	client->errorValue = drawId;
	*error = __glXError(GLXBadDrawable);
	return NULL;
    }

    if (pDraw->pScreen != glxc->pGlxScreen->pScreen) {
	client->errorValue = pDraw->pScreen->myNum;
	*error = BadMatch;
	return NULL;
    }

    if (!validGlxFBConfigForWindow(client, glxc->config, pDraw, error))
	return NULL;

    pGlxDraw = glxc->pGlxScreen->createDrawable(client, glxc->pGlxScreen,
						pDraw, drawId,
						GLX_DRAWABLE_WINDOW,
						drawId, glxc->config);

    /* since we are creating the drawablePrivate, drawId should be new */
    if (!AddResource(drawId, __glXDrawableRes, pGlxDraw)) {
	pGlxDraw->destroy (pGlxDraw);
	*error = BadAlloc;
	return NULL;
    }

    return pGlxDraw;
}

/*****************************************************************************/
/*
** Make an OpenGL context and drawable current.
*/

static int
DoMakeCurrent(__GLXclientState *cl,
	      GLXDrawable drawId, GLXDrawable readId,
	      GLXContextID contextId, GLXContextTag tag)
{
    ClientPtr client = cl->client;
    xGLXMakeCurrentReply reply;
    __GLXcontext *glxc, *prevglxc;
    __GLXdrawable *drawPriv = NULL;
    __GLXdrawable *readPriv = NULL;
    int error;
    GLuint  mask;

    /*
    ** If one is None and the other isn't, it's a bad match.
    */

    mask  = (drawId == None)    ? (1 << 0) : 0;
    mask |= (readId == None)    ? (1 << 1) : 0;
    mask |= (contextId == None) ? (1 << 2) : 0;

    if ( (mask != 0x00) && (mask != 0x07) ) {
	return BadMatch;
    }
    
    /*
    ** Lookup old context.  If we have one, it must be in a usable state.
    */
    if (tag != 0) {
	prevglxc = __glXLookupContextByTag(cl, tag);
	if (!prevglxc) {
	    /*
	    ** Tag for previous context is invalid.
	    */
	    return __glXError(GLXBadContextTag);
	}
	if (prevglxc->renderMode != GL_RENDER) {
	    /* Oops.  Not in render mode render. */
	    client->errorValue = prevglxc->id;
	    return __glXError(GLXBadContextState);
	}
    } else {
	prevglxc = 0;
    }

    /*
    ** Lookup new context.  It must not be current for someone else.
    */
    if (contextId != None) {
	int  status;

	if (!validGlxContext(client, contextId, DixUseAccess, &glxc, &error))
	    return error;
	if ((glxc != prevglxc) && glxc->isCurrent) {
	    /* Context is current to somebody else */
	    return BadAccess;
	}

	assert( drawId != None );
	assert( readId != None );

	drawPriv = __glXGetDrawable(glxc, drawId, client, &status);
	if (drawPriv == NULL)
	    return status;

	readPriv = __glXGetDrawable(glxc, readId, client, &status);
	if (readPriv == NULL)
	    return status;

    } else {
	/* Switching to no context.  Ignore new drawable. */
	glxc = 0;
	drawPriv = 0;
	readPriv = 0;
    }


    if (prevglxc) {
	/*
	** Flush the previous context if needed.
	*/
	if (__GLX_HAS_UNFLUSHED_CMDS(prevglxc)) {
	    if (__glXForceCurrent(cl, tag, (int *)&error)) {
		CALL_Flush( GET_DISPATCH(), () );
		__GLX_NOTE_FLUSHED_CMDS(prevglxc);
	    } else {
		return error;
	    }
	}

	/*
	** Make the previous context not current.
	*/
	if (!(*prevglxc->loseCurrent)(prevglxc)) {
	    return __glXError(GLXBadContext);
	}
	__glXFlushContextCache();
	if (!prevglxc->isDirect) {
	    prevglxc->drawPriv = NULL;
	    prevglxc->readPriv = NULL;
	}
    }
	

    if ((glxc != 0) && !glxc->isDirect) {

	glxc->drawPriv = drawPriv;
	glxc->readPriv = readPriv;

	/* make the context current */
	if (!(*glxc->makeCurrent)(glxc)) {
	    glxc->drawPriv = NULL;
	    glxc->readPriv = NULL;
	    return __glXError(GLXBadContext);
	}

	glxc->isCurrent = GL_TRUE;
    }

    StopUsingContext(prevglxc);

    if (glxc) {
	StartUsingContext(cl, glxc);
	reply.contextTag = glxc->id;
    } else {
	reply.contextTag = 0;
    }

    reply.length = 0;
    reply.type = X_Reply;
    reply.sequenceNumber = client->sequence;

    if (client->swapped) {
	__glXSwapMakeCurrentReply(client, &reply);
    } else {
	WriteToClient(client, sz_xGLXMakeCurrentReply, (char *)&reply);
    }
    return Success;
}

int __glXDisp_MakeCurrent(__GLXclientState *cl, GLbyte *pc)
{
    ClientPtr client = cl->client;
    xGLXMakeCurrentReq *req = (xGLXMakeCurrentReq *) pc;

    REQUEST_SIZE_MATCH(xGLXMakeCurrentReq);

    return DoMakeCurrent( cl, req->drawable, req->drawable,
			  req->context, req->oldContextTag );
}

int __glXDisp_MakeContextCurrent(__GLXclientState *cl, GLbyte *pc)
{
    ClientPtr client = cl->client;
    xGLXMakeContextCurrentReq *req = (xGLXMakeContextCurrentReq *) pc;

    REQUEST_SIZE_MATCH(xGLXMakeContextCurrentReq);

    return DoMakeCurrent( cl, req->drawable, req->readdrawable,
			  req->context, req->oldContextTag );
}

int __glXDisp_MakeCurrentReadSGI(__GLXclientState *cl, GLbyte *pc)
{
    ClientPtr client = cl->client;
    xGLXMakeCurrentReadSGIReq *req = (xGLXMakeCurrentReadSGIReq *) pc;

    REQUEST_SIZE_MATCH(xGLXMakeCurrentReadSGIReq);

    return DoMakeCurrent( cl, req->drawable, req->readable,
			  req->context, req->oldContextTag );
}

int __glXDisp_IsDirect(__GLXclientState *cl, GLbyte *pc)
{
    ClientPtr client = cl->client;
    xGLXIsDirectReq *req = (xGLXIsDirectReq *) pc;
    xGLXIsDirectReply reply;
    __GLXcontext *glxc;
    int err;

    REQUEST_SIZE_MATCH(xGLXIsDirectReq);

    if (!validGlxContext(cl->client, req->context, DixReadAccess, &glxc, &err))
	return err;

    reply.isDirect = glxc->isDirect;
    reply.length = 0;
    reply.type = X_Reply;
    reply.sequenceNumber = client->sequence;

    if (client->swapped) {
	__glXSwapIsDirectReply(client, &reply);
    } else {
	WriteToClient(client, sz_xGLXIsDirectReply, (char *)&reply);
    }

    return Success;
}

int __glXDisp_QueryVersion(__GLXclientState *cl, GLbyte *pc)
{
    ClientPtr client = cl->client;
    xGLXQueryVersionReq *req = (xGLXQueryVersionReq *) pc;
    xGLXQueryVersionReply reply;
    GLuint major, minor;

    REQUEST_SIZE_MATCH(xGLXQueryVersionReq);

    major = req->majorVersion;
    minor = req->minorVersion;
    (void)major;
    (void)minor;

    /*
    ** Server should take into consideration the version numbers sent by the
    ** client if it wants to work with older clients; however, in this
    ** implementation the server just returns its version number.
    */
    reply.majorVersion = glxMajorVersion;
    reply.minorVersion = glxMinorVersion;
    reply.length = 0;
    reply.type = X_Reply;
    reply.sequenceNumber = client->sequence;

    if (client->swapped) {
	__glXSwapQueryVersionReply(client, &reply);
    } else {
	WriteToClient(client, sz_xGLXQueryVersionReply, (char *)&reply);
    }
    return Success;
}

int __glXDisp_WaitGL(__GLXclientState *cl, GLbyte *pc)
{
    ClientPtr client = cl->client;
    xGLXWaitGLReq *req = (xGLXWaitGLReq *)pc;
    GLXContextTag tag;
    __GLXcontext *glxc = NULL;
    int error;

    REQUEST_SIZE_MATCH(xGLXWaitGLReq);

    tag = req->contextTag;
    if (tag) {
	glxc = __glXLookupContextByTag(cl, tag);
	if (!glxc)
	    return __glXError(GLXBadContextTag);
    
	if (!__glXForceCurrent(cl, req->contextTag, &error))
	    return error;

	CALL_Finish( GET_DISPATCH(), () );
    }

    if (glxc && glxc->drawPriv->waitGL)
	(*glxc->drawPriv->waitGL)(glxc->drawPriv);

    return Success;
}

int __glXDisp_WaitX(__GLXclientState *cl, GLbyte *pc)
{
    ClientPtr client = cl->client;
    xGLXWaitXReq *req = (xGLXWaitXReq *)pc;
    GLXContextTag tag;
    __GLXcontext *glxc = NULL;
    int error;

    REQUEST_SIZE_MATCH(xGLXWaitXReq);

    tag = req->contextTag;
    if (tag) {
	glxc = __glXLookupContextByTag(cl, tag);
	if (!glxc)
	    return __glXError(GLXBadContextTag);
    
	if (!__glXForceCurrent(cl, req->contextTag, &error))
	    return error;
    }

    if (glxc && glxc->drawPriv->waitX)
	(*glxc->drawPriv->waitX)(glxc->drawPriv);

    return Success;
}

int __glXDisp_CopyContext(__GLXclientState *cl, GLbyte *pc)
{
    ClientPtr client = cl->client;
    xGLXCopyContextReq *req = (xGLXCopyContextReq *) pc;
    GLXContextID source;
    GLXContextID dest;
    GLXContextTag tag;
    unsigned long mask;
    __GLXcontext *src, *dst;
    int error;

    REQUEST_SIZE_MATCH(xGLXCopyContextReq);

    source = req->source;
    dest = req->dest;
    tag = req->contextTag;
    mask = req->mask;
    if (!validGlxContext(cl->client, source, DixReadAccess, &src, &error))
	return error;
    if (!validGlxContext(cl->client, dest, DixWriteAccess, &dst, &error))
	return error;

    /*
    ** They must be in the same address space, and same screen.
    ** NOTE: no support for direct rendering contexts here.
    */
    if (src->isDirect || dst->isDirect ||
	(src->pGlxScreen != dst->pGlxScreen)) {
	client->errorValue = source;
	return BadMatch;
    }

    /*
    ** The destination context must not be current for any client.
    */
    if (dst->isCurrent) {
	client->errorValue = dest;
	return BadAccess;
    }

    if (tag) {
	__GLXcontext *tagcx = __glXLookupContextByTag(cl, tag);
	
	if (!tagcx) {
	    return __glXError(GLXBadContextTag);
	}
	if (tagcx != src) {
	    /*
	    ** This would be caused by a faulty implementation of the client
	    ** library.
	    */
	    return BadMatch;
	}
	/*
	** In this case, glXCopyContext is in both GL and X streams, in terms
	** of sequentiality.
	*/
	if (__glXForceCurrent(cl, tag, &error)) {
	    /*
	    ** Do whatever is needed to make sure that all preceding requests
	    ** in both streams are completed before the copy is executed.
	    */
	    CALL_Finish( GET_DISPATCH(), () );
	    __GLX_NOTE_FLUSHED_CMDS(tagcx);
	} else {
	    return error;
	}
    }
    /*
    ** Issue copy.  The only reason for failure is a bad mask.
    */
    if (!(*dst->copy)(dst, src, mask)) {
	client->errorValue = mask;
	return BadValue;
    }
    return Success;
}

enum {
    GLX_VIS_CONFIG_UNPAIRED = 18,
    GLX_VIS_CONFIG_PAIRED = 20
};

enum {
    GLX_VIS_CONFIG_TOTAL = GLX_VIS_CONFIG_UNPAIRED + GLX_VIS_CONFIG_PAIRED
};

int __glXDisp_GetVisualConfigs(__GLXclientState *cl, GLbyte *pc)
{
    xGLXGetVisualConfigsReq *req = (xGLXGetVisualConfigsReq *) pc;
    ClientPtr client = cl->client;
    xGLXGetVisualConfigsReply reply;
    __GLXscreen *pGlxScreen;
    __GLXconfig *modes;
    CARD32 buf[GLX_VIS_CONFIG_TOTAL];
    int p, i, err;
    __GLX_DECLARE_SWAP_VARIABLES;
    __GLX_DECLARE_SWAP_ARRAY_VARIABLES;

    REQUEST_SIZE_MATCH(xGLXGetVisualConfigsReq);

    if (!validGlxScreen(cl->client, req->screen, &pGlxScreen, &err))
	return err;

    reply.numVisuals = pGlxScreen->numVisuals;
    reply.numProps = GLX_VIS_CONFIG_TOTAL;
    reply.length = (reply.numVisuals * __GLX_SIZE_CARD32 * GLX_VIS_CONFIG_TOTAL) >> 2;
    reply.type = X_Reply;
    reply.sequenceNumber = client->sequence;

    if (client->swapped) {
	__GLX_SWAP_SHORT(&reply.sequenceNumber);
	__GLX_SWAP_INT(&reply.length);
	__GLX_SWAP_INT(&reply.numVisuals);
	__GLX_SWAP_INT(&reply.numProps);
    }

    WriteToClient(client, sz_xGLXGetVisualConfigsReply, (char *)&reply);

    for (i = 0; i < pGlxScreen->numVisuals; i++) {
	modes = pGlxScreen->visuals[i];

	p = 0;
	buf[p++] = modes->visualID;
	buf[p++] = glxConvertToXVisualType( modes->visualType );
	buf[p++] = (modes->renderType & GLX_RGBA_BIT) ? GL_TRUE : GL_FALSE;

	buf[p++] = modes->redBits;
	buf[p++] = modes->greenBits;
	buf[p++] = modes->blueBits;
	buf[p++] = modes->alphaBits;
	buf[p++] = modes->accumRedBits;
	buf[p++] = modes->accumGreenBits;
	buf[p++] = modes->accumBlueBits;
	buf[p++] = modes->accumAlphaBits;

	buf[p++] = modes->doubleBufferMode;
	buf[p++] = modes->stereoMode;

	buf[p++] = modes->rgbBits;
	buf[p++] = modes->depthBits;
	buf[p++] = modes->stencilBits;
	buf[p++] = modes->numAuxBuffers;
	buf[p++] = modes->level;

	assert(p == GLX_VIS_CONFIG_UNPAIRED);
	/* 
	** Add token/value pairs for extensions.
	*/
	buf[p++] = GLX_VISUAL_CAVEAT_EXT;
	buf[p++] = modes->visualRating;
	buf[p++] = GLX_TRANSPARENT_TYPE;
	buf[p++] = modes->transparentPixel;
	buf[p++] = GLX_TRANSPARENT_RED_VALUE;
	buf[p++] = modes->transparentRed;
	buf[p++] = GLX_TRANSPARENT_GREEN_VALUE;
	buf[p++] = modes->transparentGreen;
	buf[p++] = GLX_TRANSPARENT_BLUE_VALUE;
	buf[p++] = modes->transparentBlue;
	buf[p++] = GLX_TRANSPARENT_ALPHA_VALUE;
	buf[p++] = modes->transparentAlpha;
	buf[p++] = GLX_TRANSPARENT_INDEX_VALUE;
	buf[p++] = modes->transparentIndex;
	buf[p++] = GLX_SAMPLES_SGIS;
	buf[p++] = modes->samples;
	buf[p++] = GLX_SAMPLE_BUFFERS_SGIS;
	buf[p++] = modes->sampleBuffers;
	buf[p++] = 0; /* copy over visualSelectGroup (GLX_VISUAL_SELECT_GROUP_SGIX)? */
	buf[p++] = 0;

	assert(p == GLX_VIS_CONFIG_TOTAL);
	if (client->swapped) {
	    __GLX_SWAP_INT_ARRAY(buf, p);
	}
	WriteToClient(client, __GLX_SIZE_CARD32 * p, (char *)buf);
    }
    return Success;
}

#define __GLX_TOTAL_FBCONFIG_ATTRIBS (36)
#define __GLX_FBCONFIG_ATTRIBS_LENGTH (__GLX_TOTAL_FBCONFIG_ATTRIBS * 2)
/**
 * Send the set of GLXFBConfigs to the client.  There is not currently
 * and interface into the driver on the server-side to get GLXFBConfigs,
 * so we "invent" some based on the \c __GLXvisualConfig structures that
 * the driver does supply.
 * 
 * The reply format for both \c glXGetFBConfigs and \c glXGetFBConfigsSGIX
 * is the same, so this routine pulls double duty.
 */

static int
DoGetFBConfigs(__GLXclientState *cl, unsigned screen)
{
    ClientPtr client = cl->client;
    xGLXGetFBConfigsReply reply;
    __GLXscreen *pGlxScreen;
    CARD32 buf[__GLX_FBCONFIG_ATTRIBS_LENGTH];
    int p, err;
    __GLXconfig *modes;
    __GLX_DECLARE_SWAP_VARIABLES;
    __GLX_DECLARE_SWAP_ARRAY_VARIABLES;

    if (!validGlxScreen(cl->client, screen, &pGlxScreen, &err))
	return err;

    reply.numFBConfigs = pGlxScreen->numFBConfigs;
    reply.numAttribs = __GLX_TOTAL_FBCONFIG_ATTRIBS;
    reply.length = (__GLX_FBCONFIG_ATTRIBS_LENGTH * reply.numFBConfigs);
    reply.type = X_Reply;
    reply.sequenceNumber = client->sequence;

    if (client->swapped) {
	__GLX_SWAP_SHORT(&reply.sequenceNumber);
	__GLX_SWAP_INT(&reply.length);
	__GLX_SWAP_INT(&reply.numFBConfigs);
	__GLX_SWAP_INT(&reply.numAttribs);
    }

    WriteToClient(client, sz_xGLXGetFBConfigsReply, (char *)&reply);

    for (modes = pGlxScreen->fbconfigs; modes != NULL; modes = modes->next) {
	p = 0;

#define WRITE_PAIR(tag,value) \
    do { buf[p++] = tag ; buf[p++] = value ; } while( 0 )

	WRITE_PAIR( GLX_VISUAL_ID,        modes->visualID );
	WRITE_PAIR( GLX_FBCONFIG_ID,      modes->fbconfigID );
	WRITE_PAIR( GLX_X_RENDERABLE,     GL_TRUE );

	WRITE_PAIR( GLX_RGBA,
		    (modes->renderType & GLX_RGBA_BIT) ? GL_TRUE : GL_FALSE );
	WRITE_PAIR( GLX_RENDER_TYPE,      modes->renderType );
	WRITE_PAIR( GLX_DOUBLEBUFFER,     modes->doubleBufferMode );
	WRITE_PAIR( GLX_STEREO,           modes->stereoMode );

	WRITE_PAIR( GLX_BUFFER_SIZE,      modes->rgbBits );
	WRITE_PAIR( GLX_LEVEL,            modes->level );
	WRITE_PAIR( GLX_AUX_BUFFERS,      modes->numAuxBuffers );
	WRITE_PAIR( GLX_RED_SIZE,         modes->redBits );
	WRITE_PAIR( GLX_GREEN_SIZE,       modes->greenBits );
	WRITE_PAIR( GLX_BLUE_SIZE,        modes->blueBits );
	WRITE_PAIR( GLX_ALPHA_SIZE,       modes->alphaBits );
	WRITE_PAIR( GLX_ACCUM_RED_SIZE,   modes->accumRedBits );
	WRITE_PAIR( GLX_ACCUM_GREEN_SIZE, modes->accumGreenBits );
	WRITE_PAIR( GLX_ACCUM_BLUE_SIZE,  modes->accumBlueBits );
	WRITE_PAIR( GLX_ACCUM_ALPHA_SIZE, modes->accumAlphaBits );
	WRITE_PAIR( GLX_DEPTH_SIZE,       modes->depthBits );
	WRITE_PAIR( GLX_STENCIL_SIZE,     modes->stencilBits );
	WRITE_PAIR( GLX_X_VISUAL_TYPE,    modes->visualType );
	WRITE_PAIR( GLX_CONFIG_CAVEAT, modes->visualRating );
	WRITE_PAIR( GLX_TRANSPARENT_TYPE, modes->transparentPixel );
	WRITE_PAIR( GLX_TRANSPARENT_RED_VALUE, modes->transparentRed );
	WRITE_PAIR( GLX_TRANSPARENT_GREEN_VALUE, modes->transparentGreen );
	WRITE_PAIR( GLX_TRANSPARENT_BLUE_VALUE, modes->transparentBlue );
	WRITE_PAIR( GLX_TRANSPARENT_ALPHA_VALUE, modes->transparentAlpha );
	WRITE_PAIR( GLX_TRANSPARENT_INDEX_VALUE, modes->transparentIndex );
	WRITE_PAIR( GLX_SWAP_METHOD_OML, modes->swapMethod );
	WRITE_PAIR( GLX_SAMPLES_SGIS, modes->samples );
	WRITE_PAIR( GLX_SAMPLE_BUFFERS_SGIS, modes->sampleBuffers );
	/* GLX_VISUAL_SELECT_GROUP_SGIX ? */
	WRITE_PAIR( GLX_DRAWABLE_TYPE, modes->drawableType );
	WRITE_PAIR( GLX_BIND_TO_TEXTURE_RGB_EXT, modes->bindToTextureRgb );
	WRITE_PAIR( GLX_BIND_TO_TEXTURE_RGBA_EXT, modes->bindToTextureRgba );
	WRITE_PAIR( GLX_BIND_TO_MIPMAP_TEXTURE_EXT, modes->bindToMipmapTexture );
	WRITE_PAIR( GLX_BIND_TO_TEXTURE_TARGETS_EXT, modes->bindToTextureTargets );

	if (client->swapped) {
	    __GLX_SWAP_INT_ARRAY(buf, __GLX_FBCONFIG_ATTRIBS_LENGTH);
	}
	WriteToClient(client, __GLX_SIZE_CARD32 * __GLX_FBCONFIG_ATTRIBS_LENGTH,
		      (char *)buf);
    }
    return Success;
}


int __glXDisp_GetFBConfigs(__GLXclientState *cl, GLbyte *pc)
{
    ClientPtr client = cl->client;
    xGLXGetFBConfigsReq *req = (xGLXGetFBConfigsReq *) pc;
    REQUEST_SIZE_MATCH(xGLXGetFBConfigsReq);
    return DoGetFBConfigs(cl, req->screen);
}

int __glXDisp_GetFBConfigsSGIX(__GLXclientState *cl, GLbyte *pc)
{
    ClientPtr client = cl->client;
    xGLXGetFBConfigsSGIXReq *req = (xGLXGetFBConfigsSGIXReq *) pc;
    /* work around mesa bug, don't use REQUEST_SIZE_MATCH */
    REQUEST_AT_LEAST_SIZE(xGLXGetFBConfigsSGIXReq);
    return DoGetFBConfigs(cl, req->screen);
}

GLboolean
__glXDrawableInit(__GLXdrawable *drawable,
		  __GLXscreen *screen, DrawablePtr pDraw, int type,
		  XID drawId, __GLXconfig *config)
{
    drawable->pDraw = pDraw;
    drawable->type = type;
    drawable->drawId = drawId;
    drawable->config = config;
    drawable->eventMask = 0;

    return GL_TRUE;
}

void
__glXDrawableRelease(__GLXdrawable *drawable)
{
}

static int 
DoCreateGLXDrawable(ClientPtr client, __GLXscreen *pGlxScreen,
		    __GLXconfig *config, DrawablePtr pDraw, XID drawableId,
		    XID glxDrawableId, int type)
{
    __GLXdrawable *pGlxDraw;

    if (pGlxScreen->pScreen != pDraw->pScreen)
	return BadMatch;

    pGlxDraw = pGlxScreen->createDrawable(client, pGlxScreen, pDraw,
					  drawableId, type,
					  glxDrawableId, config);
    if (pGlxDraw == NULL)
	return BadAlloc;

    if (!AddResource(glxDrawableId, __glXDrawableRes, pGlxDraw)) {
	pGlxDraw->destroy (pGlxDraw);
	return BadAlloc;
    }

    /*
     * Windows aren't refcounted, so track both the X and the GLX window
     * so we get called regardless of destruction order.
     */
    if (drawableId != glxDrawableId && type == GLX_DRAWABLE_WINDOW &&
	!AddResource(pDraw->id, __glXDrawableRes, pGlxDraw)) {
	pGlxDraw->destroy (pGlxDraw);
	return BadAlloc;
    }

    return Success;
}

static int
DoCreateGLXPixmap(ClientPtr client, __GLXscreen *pGlxScreen, __GLXconfig *config,
		  XID drawableId, XID glxDrawableId)
{
    DrawablePtr pDraw;
    int err;

    LEGAL_NEW_RESOURCE(glxDrawableId, client);

    err = dixLookupDrawable(&pDraw, drawableId, client, 0, DixAddAccess);
    if (err != Success) {
	client->errorValue = drawableId;
	return err;
    }
    if (pDraw->type != DRAWABLE_PIXMAP) {
	client->errorValue = drawableId;
	return BadPixmap;
    }

    err = DoCreateGLXDrawable(client, pGlxScreen, config, pDraw, drawableId,
			      glxDrawableId, GLX_DRAWABLE_PIXMAP);

    ((PixmapPtr)pDraw)->refcnt++;

    return err;
}

static void
determineTextureTarget(ClientPtr client, XID glxDrawableID,
		       CARD32 *attribs, CARD32 numAttribs)
{
    GLenum target = 0;
    GLenum format = 0;
    int i, err;
    __GLXdrawable *pGlxDraw;

    if (!validGlxDrawable(client, glxDrawableID, GLX_DRAWABLE_PIXMAP,
			  DixWriteAccess, &pGlxDraw, &err))
	/* We just added it in CreatePixmap, so we should never get here. */
	return;

    for (i = 0; i < numAttribs; i++) {
	if (attribs[2 * i] == GLX_TEXTURE_TARGET_EXT) {
	    switch (attribs[2 * i + 1]) {
	    case GLX_TEXTURE_2D_EXT:
		target = GL_TEXTURE_2D;
		break;
	    case GLX_TEXTURE_RECTANGLE_EXT:
		target = GL_TEXTURE_RECTANGLE_ARB;
		break;
	    }
	}

	if (attribs[2 * i] == GLX_TEXTURE_FORMAT_EXT)
		format = attribs[2 * i + 1];
    }
 
    if (!target) {
	int w = pGlxDraw->pDraw->width, h = pGlxDraw->pDraw->height;
	
	if (h & (h - 1) || w & (w - 1))
	    target = GL_TEXTURE_RECTANGLE_ARB;
	else
	    target = GL_TEXTURE_2D;
    }

    pGlxDraw->target = target;
    pGlxDraw->format = format;
}

int __glXDisp_CreateGLXPixmap(__GLXclientState *cl, GLbyte *pc)
{
    ClientPtr client = cl->client;
    xGLXCreateGLXPixmapReq *req = (xGLXCreateGLXPixmapReq *) pc;
    __GLXconfig *config;
    __GLXscreen *pGlxScreen;
    int err;

    REQUEST_SIZE_MATCH(xGLXCreateGLXPixmapReq);

    if (!validGlxScreen(cl->client, req->screen, &pGlxScreen, &err))
	return err;
    if (!validGlxVisual(cl->client, pGlxScreen, req->visual, &config, &err))
	return err;

    return DoCreateGLXPixmap(cl->client, pGlxScreen, config,
			     req->pixmap, req->glxpixmap);
}

int __glXDisp_CreatePixmap(__GLXclientState *cl, GLbyte *pc)
{
    ClientPtr client = cl->client;
    xGLXCreatePixmapReq *req = (xGLXCreatePixmapReq *) pc;
    __GLXconfig *config;
    __GLXscreen *pGlxScreen;
    int err;

    REQUEST_AT_LEAST_SIZE(xGLXCreatePixmapReq);
    if (req->numAttribs > (UINT32_MAX >> 3)) {
	client->errorValue = req->numAttribs;
	return BadValue;
    }
    REQUEST_FIXED_SIZE(xGLXCreatePixmapReq, req->numAttribs << 3);

    if (!validGlxScreen(cl->client, req->screen, &pGlxScreen, &err))
	return err;
    if (!validGlxFBConfig(cl->client, pGlxScreen, req->fbconfig, &config, &err))
	return err;

    err = DoCreateGLXPixmap(cl->client, pGlxScreen, config,
			    req->pixmap, req->glxpixmap);
    if (err != Success)
	return err;

    determineTextureTarget(cl->client, req->glxpixmap,
			   (CARD32*) (req + 1), req->numAttribs);

    return Success;
}

int __glXDisp_CreateGLXPixmapWithConfigSGIX(__GLXclientState *cl, GLbyte *pc)
{
    ClientPtr client = cl->client;
    xGLXCreateGLXPixmapWithConfigSGIXReq *req = 
	(xGLXCreateGLXPixmapWithConfigSGIXReq *) pc;
    __GLXconfig *config;
    __GLXscreen *pGlxScreen;
    int err;

    REQUEST_SIZE_MATCH(xGLXCreateGLXPixmapWithConfigSGIXReq);

    if (!validGlxScreen(cl->client, req->screen, &pGlxScreen, &err))
	return err;
    if (!validGlxFBConfig(cl->client, pGlxScreen, req->fbconfig, &config, &err))
	return err;

    return DoCreateGLXPixmap(cl->client, pGlxScreen,
			     config, req->pixmap, req->glxpixmap);
}


static int DoDestroyDrawable(__GLXclientState *cl, XID glxdrawable, int type)
{
    __GLXdrawable *pGlxDraw;
    int err;

    if (!validGlxDrawable(cl->client, glxdrawable, type,
			  DixDestroyAccess, &pGlxDraw, &err))
	return err;

    FreeResource(glxdrawable, FALSE);

    return Success;
}

int __glXDisp_DestroyGLXPixmap(__GLXclientState *cl, GLbyte *pc)
{
    ClientPtr client = cl->client;
    xGLXDestroyGLXPixmapReq *req = (xGLXDestroyGLXPixmapReq *) pc;

    REQUEST_SIZE_MATCH(xGLXDestroyGLXPixmapReq);

    return DoDestroyDrawable(cl, req->glxpixmap, GLX_DRAWABLE_PIXMAP);
}

int __glXDisp_DestroyPixmap(__GLXclientState *cl, GLbyte *pc)
{
    ClientPtr client = cl->client;
    xGLXDestroyPixmapReq *req = (xGLXDestroyPixmapReq *) pc;

    /* should be REQUEST_SIZE_MATCH, but mesa's glXDestroyPixmap used to set
     * length to 3 instead of 2 */
    REQUEST_AT_LEAST_SIZE(xGLXDestroyPixmapReq);

    return DoDestroyDrawable(cl, req->glxpixmap, GLX_DRAWABLE_PIXMAP);
}

static int
DoCreatePbuffer(ClientPtr client, int screenNum, XID fbconfigId,
		int width, int height, XID glxDrawableId)
{
    __GLXconfig	*config;
    __GLXscreen		*pGlxScreen;
    PixmapPtr		 pPixmap;
    int			 err;

    LEGAL_NEW_RESOURCE(glxDrawableId, client);

    if (!validGlxScreen(client, screenNum, &pGlxScreen, &err))
	return err;
    if (!validGlxFBConfig(client, pGlxScreen, fbconfigId, &config, &err))
	return err;

    __glXenterServer(GL_FALSE);
    pPixmap = (*pGlxScreen->pScreen->CreatePixmap) (pGlxScreen->pScreen,
						    width, height, config->rgbBits, 0);
    __glXleaveServer(GL_FALSE);

    /* Assign the pixmap the same id as the pbuffer and add it as a
     * resource so it and the DRI2 drawable will be reclaimed when the
     * pbuffer is destroyed. */
    pPixmap->drawable.id = glxDrawableId;
    if (!AddResource(pPixmap->drawable.id, RT_PIXMAP, pPixmap))
	return BadAlloc;

    return DoCreateGLXDrawable(client, pGlxScreen, config, &pPixmap->drawable,
			       glxDrawableId, glxDrawableId,
			       GLX_DRAWABLE_PBUFFER);
}

int __glXDisp_CreatePbuffer(__GLXclientState *cl, GLbyte *pc)
{
    ClientPtr client = cl->client;
    xGLXCreatePbufferReq	*req = (xGLXCreatePbufferReq *) pc;
    CARD32			*attrs;
    int				 width, height, i;

    REQUEST_AT_LEAST_SIZE(xGLXCreatePbufferReq);
    if (req->numAttribs > (UINT32_MAX >> 3)) {
	client->errorValue = req->numAttribs;
	return BadValue;
    }
    REQUEST_FIXED_SIZE(xGLXCreatePbufferReq, req->numAttribs << 3);

    attrs = (CARD32 *) (req + 1);
    width = 0;
    height = 0;

    for (i = 0; i < req->numAttribs; i++) {
	switch (attrs[i * 2]) {
	case GLX_PBUFFER_WIDTH:
	    width = attrs[i * 2 + 1];
	    break;
	case GLX_PBUFFER_HEIGHT:
	    height = attrs[i * 2 + 1];
	    break;
	case GLX_LARGEST_PBUFFER:
	case GLX_PRESERVED_CONTENTS:
	    /* FIXME: huh... */
	    break;
	}
    }

    return DoCreatePbuffer(cl->client, req->screen, req->fbconfig,
			   width, height, req->pbuffer);
}

int __glXDisp_CreateGLXPbufferSGIX(__GLXclientState *cl, GLbyte *pc)
{
    ClientPtr client = cl->client;
    xGLXCreateGLXPbufferSGIXReq *req = (xGLXCreateGLXPbufferSGIXReq *) pc;

    REQUEST_AT_LEAST_SIZE(xGLXCreateGLXPbufferSGIXReq);

    return DoCreatePbuffer(cl->client, req->screen, req->fbconfig,
			   req->width, req->height, req->pbuffer);
}

int __glXDisp_DestroyPbuffer(__GLXclientState *cl, GLbyte *pc)
{
    ClientPtr client = cl->client;
    xGLXDestroyPbufferReq *req = (xGLXDestroyPbufferReq *) pc;

    REQUEST_SIZE_MATCH(xGLXDestroyPbufferReq);

    return DoDestroyDrawable(cl, req->pbuffer, GLX_DRAWABLE_PBUFFER);
}

int __glXDisp_DestroyGLXPbufferSGIX(__GLXclientState *cl, GLbyte *pc)
{
    ClientPtr client = cl->client;
    xGLXDestroyGLXPbufferSGIXReq *req = (xGLXDestroyGLXPbufferSGIXReq *) pc;

    REQUEST_SIZE_MATCH(xGLXDestroyGLXPbufferSGIXReq);

    return DoDestroyDrawable(cl, req->pbuffer, GLX_DRAWABLE_PBUFFER);
}

static int
DoChangeDrawableAttributes(ClientPtr client, XID glxdrawable,
			   int numAttribs, CARD32 *attribs)
{
    __GLXdrawable *pGlxDraw;
    int i, err;

    if (!validGlxDrawable(client, glxdrawable, GLX_DRAWABLE_ANY,
			  DixSetAttrAccess, &pGlxDraw, &err))
	return err;

    for (i = 0; i < numAttribs; i++) {
	switch(attribs[i * 2]) {
	case GLX_EVENT_MASK:
	    /* All we do is to record the event mask so we can send it
	     * back when queried.  We never actually clobber the
	     * pbuffers, so we never need to send out the event. */
	    pGlxDraw->eventMask = attribs[i * 2 + 1];
	    break;
	}
    }

    return Success;
}

int __glXDisp_ChangeDrawableAttributes(__GLXclientState *cl, GLbyte *pc)
{
    ClientPtr client = cl->client;
    xGLXChangeDrawableAttributesReq *req =
	(xGLXChangeDrawableAttributesReq *) pc;

    REQUEST_AT_LEAST_SIZE(xGLXChangeDrawableAttributesReq);
    if (req->numAttribs > (UINT32_MAX >> 3)) {
	client->errorValue = req->numAttribs;
	return BadValue;
    }
#if 0
    /* mesa sends an additional 8 bytes */
    REQUEST_FIXED_SIZE(xGLXChangeDrawableAttributesReq, req->numAttribs << 3);
#else
    if (((sizeof(xGLXChangeDrawableAttributesReq) + (req->numAttribs << 3)) >> 2) < client->req_len)
	    return BadLength;
#endif

    return DoChangeDrawableAttributes(cl->client, req->drawable,
				      req->numAttribs, (CARD32 *) (req + 1));
}

int __glXDisp_ChangeDrawableAttributesSGIX(__GLXclientState *cl, GLbyte *pc)
{
    ClientPtr client = cl->client;
    xGLXChangeDrawableAttributesSGIXReq *req =
	(xGLXChangeDrawableAttributesSGIXReq *)pc;

    REQUEST_AT_LEAST_SIZE(xGLXChangeDrawableAttributesSGIXReq);
    if (req->numAttribs > (UINT32_MAX >> 3)) {
	client->errorValue = req->numAttribs;
	return BadValue;
    }
    REQUEST_FIXED_SIZE(xGLXChangeDrawableAttributesSGIXReq, req->numAttribs << 3);

    return DoChangeDrawableAttributes(cl->client, req->drawable,
				      req->numAttribs, (CARD32 *) (req + 1));
}

int __glXDisp_CreateWindow(__GLXclientState *cl, GLbyte *pc)
{
    xGLXCreateWindowReq	*req = (xGLXCreateWindowReq *) pc;
    __GLXconfig	*config;
    __GLXscreen		*pGlxScreen;
    ClientPtr		 client = cl->client;
    DrawablePtr		 pDraw;
    int			 err;

    REQUEST_AT_LEAST_SIZE(xGLXCreateWindowReq);
    if (req->numAttribs > (UINT32_MAX >> 3)) {
	client->errorValue = req->numAttribs;
	return BadValue;
    }
    REQUEST_FIXED_SIZE(xGLXCreateWindowReq, req->numAttribs << 3);

    LEGAL_NEW_RESOURCE(req->glxwindow, client);

    if (!validGlxScreen(client, req->screen, &pGlxScreen, &err))
	return err;
    if (!validGlxFBConfig(client, pGlxScreen, req->fbconfig, &config, &err))
	return err;

    err = dixLookupDrawable(&pDraw, req->window, client, 0, DixAddAccess);
    if (err != Success || pDraw->type != DRAWABLE_WINDOW) {
	client->errorValue = req->window;
	return BadWindow;
    }

    if (!validGlxFBConfigForWindow(client, config, pDraw, &err))
	return err;

    return DoCreateGLXDrawable(client, pGlxScreen, config,
			       pDraw, req->window,
			       req->glxwindow, GLX_DRAWABLE_WINDOW);
}

int __glXDisp_DestroyWindow(__GLXclientState *cl, GLbyte *pc)
{
    ClientPtr client = cl->client;
    xGLXDestroyWindowReq *req = (xGLXDestroyWindowReq *) pc;

    /* mesa's glXDestroyWindow used to set length to 3 instead of 2 */
    REQUEST_AT_LEAST_SIZE(xGLXDestroyWindowReq);

    return DoDestroyDrawable(cl, req->glxwindow, GLX_DRAWABLE_WINDOW);
}


/*****************************************************************************/

/*
** NOTE: There is no portable implementation for swap buffers as of
** this time that is of value.  Consequently, this code must be
** implemented by somebody other than SGI.
*/
int __glXDisp_SwapBuffers(__GLXclientState *cl, GLbyte *pc)
{
    ClientPtr client = cl->client;
    xGLXSwapBuffersReq *req = (xGLXSwapBuffersReq *) pc;
    GLXContextTag tag;
    XID drawId;
    __GLXcontext *glxc = NULL;
    __GLXdrawable *pGlxDraw;
    int error;

    REQUEST_SIZE_MATCH(xGLXSwapBuffersReq);

    tag = req->contextTag;
    drawId = req->drawable;
    if (tag) {
	glxc = __glXLookupContextByTag(cl, tag);
	if (!glxc) {
	    return __glXError(GLXBadContextTag);
	}
	/*
	** The calling thread is swapping its current drawable.  In this case,
	** glxSwapBuffers is in both GL and X streams, in terms of
	** sequentiality.
	*/
	if (__glXForceCurrent(cl, tag, &error)) {
	    /*
	    ** Do whatever is needed to make sure that all preceding requests
	    ** in both streams are completed before the swap is executed.
	    */
	    CALL_Finish( GET_DISPATCH(), () );
	    __GLX_NOTE_FLUSHED_CMDS(glxc);
	} else {
	    return error;
	}
    }

    pGlxDraw = __glXGetDrawable(glxc, drawId, client, &error);
    if (pGlxDraw == NULL)
	return error;

    if (pGlxDraw->type == DRAWABLE_WINDOW &&
	(*pGlxDraw->swapBuffers)(cl->client, pGlxDraw) == GL_FALSE)
	return __glXError(GLXBadDrawable);

    return Success;
}


static int
DoQueryContext(__GLXclientState *cl, GLXContextID gcId)
{
    ClientPtr client = cl->client;
    __GLXcontext *ctx;
    xGLXQueryContextInfoEXTReply reply;
    int nProps;
    int *sendBuf, *pSendBuf;
    int nReplyBytes;
    int err;

    if (!validGlxContext(cl->client, gcId, DixReadAccess, &ctx, &err))
	return err;

    nProps = 3;
    reply.length = nProps << 1;
    reply.type = X_Reply;
    reply.sequenceNumber = client->sequence;
    reply.n = nProps;

    nReplyBytes = reply.length << 2;
    sendBuf = (int *)malloc((size_t)nReplyBytes);
    if (sendBuf == NULL) {
	return __glXError(GLXBadContext);	/* XXX: Is this correct? */
    }
    pSendBuf = sendBuf;
    *pSendBuf++ = GLX_SHARE_CONTEXT_EXT;
    *pSendBuf++ = (int)(ctx->share_id);
    *pSendBuf++ = GLX_VISUAL_ID_EXT;
    *pSendBuf++ = (int)(ctx->config->visualID);
    *pSendBuf++ = GLX_SCREEN_EXT;
    *pSendBuf++ = (int)(ctx->pGlxScreen->pScreen->myNum);

    if (client->swapped) {
	__glXSwapQueryContextInfoEXTReply(client, &reply, sendBuf);
    } else {
	WriteToClient(client, sz_xGLXQueryContextInfoEXTReply, (char *)&reply);
	WriteToClient(client, nReplyBytes, (char *)sendBuf);
    }
    free((char *)sendBuf);

    return Success;
}

int __glXDisp_QueryContextInfoEXT(__GLXclientState *cl, GLbyte *pc)
{
    ClientPtr client = cl->client;
    xGLXQueryContextInfoEXTReq *req = (xGLXQueryContextInfoEXTReq *) pc;

    REQUEST_SIZE_MATCH(xGLXQueryContextInfoEXTReq);

    return DoQueryContext(cl, req->context);
}

int __glXDisp_QueryContext(__GLXclientState *cl, GLbyte *pc)
{
    ClientPtr client = cl->client;
    xGLXQueryContextReq *req = (xGLXQueryContextReq *) pc;

    REQUEST_SIZE_MATCH(xGLXQueryContextReq);

    return DoQueryContext(cl, req->context);
}

int __glXDisp_BindTexImageEXT(__GLXclientState *cl, GLbyte *pc)
{
    xGLXVendorPrivateReq *req = (xGLXVendorPrivateReq *) pc;
    ClientPtr		 client = cl->client;
    __GLXcontext	*context;
    __GLXdrawable	*pGlxDraw;
    GLXDrawable		 drawId;
    int			 buffer;
    int			 error;
    CARD32		 num_attribs;

    if ((sizeof(xGLXVendorPrivateReq) + 12) >> 2 > client->req_len)
	return BadLength;

    pc += __GLX_VENDPRIV_HDR_SIZE;

    drawId = *((CARD32 *) (pc));
    buffer = *((INT32 *)  (pc + 4));
    num_attribs = *((CARD32 *) (pc + 8));
    if (num_attribs > (UINT32_MAX >> 3)) {
	client->errorValue = num_attribs;
	return BadValue;
    }
    REQUEST_FIXED_SIZE(xGLXVendorPrivateReq, 12 + (num_attribs << 3));

    if (buffer != GLX_FRONT_LEFT_EXT)
	return __glXError(GLXBadPixmap);

    context = __glXForceCurrent (cl, req->contextTag, &error);
    if (!context)
	return error;

    if (!validGlxDrawable(client, drawId, GLX_DRAWABLE_PIXMAP,
			  DixReadAccess, &pGlxDraw, &error))
	return error;

    if (!context->textureFromPixmap)
	return __glXError(GLXUnsupportedPrivateRequest);

    return context->textureFromPixmap->bindTexImage(context,
						    buffer,
						    pGlxDraw);
}

int __glXDisp_ReleaseTexImageEXT(__GLXclientState *cl, GLbyte *pc)
{
    xGLXVendorPrivateReq *req = (xGLXVendorPrivateReq *) pc;
    ClientPtr		 client = cl->client;
    __GLXdrawable	*pGlxDraw;
    __GLXcontext	*context;
    GLXDrawable		 drawId;
    int			 buffer;
    int			 error;

    REQUEST_FIXED_SIZE(xGLXVendorPrivateReq, 8);

    pc += __GLX_VENDPRIV_HDR_SIZE;

    drawId = *((CARD32 *) (pc));
    buffer = *((INT32 *)  (pc + 4));
    
    context = __glXForceCurrent (cl, req->contextTag, &error);
    if (!context)
	return error;

    if (!validGlxDrawable(client, drawId, GLX_DRAWABLE_PIXMAP,
			  DixReadAccess, &pGlxDraw, &error))
	return error;

    if (!context->textureFromPixmap)
	return __glXError(GLXUnsupportedPrivateRequest);

    return context->textureFromPixmap->releaseTexImage(context,
						       buffer,
						       pGlxDraw);
}

int __glXDisp_CopySubBufferMESA(__GLXclientState *cl, GLbyte *pc)
{
    xGLXVendorPrivateReq *req = (xGLXVendorPrivateReq *) pc;
    GLXContextTag         tag = req->contextTag;
    __GLXcontext         *glxc = NULL;
    __GLXdrawable        *pGlxDraw;
    ClientPtr		  client = cl->client;
    GLXDrawable		  drawId;
    int                   error;
    int                   x, y, width, height;

    (void) client;
    (void) req;

    REQUEST_FIXED_SIZE(xGLXVendorPrivateReq, 20);

    pc += __GLX_VENDPRIV_HDR_SIZE;

    drawId = *((CARD32 *) (pc));
    x      = *((INT32 *)  (pc + 4));
    y      = *((INT32 *)  (pc + 8));
    width  = *((INT32 *)  (pc + 12));
    height = *((INT32 *)  (pc + 16));

    if (tag) {
	glxc = __glXLookupContextByTag(cl, tag);
	if (!glxc) {
	    return __glXError(GLXBadContextTag);
	}
	/*
	** The calling thread is swapping its current drawable.  In this case,
	** glxSwapBuffers is in both GL and X streams, in terms of
	** sequentiality.
	*/
	if (__glXForceCurrent(cl, tag, &error)) {
	    /*
	    ** Do whatever is needed to make sure that all preceding requests
	    ** in both streams are completed before the swap is executed.
	    */
	    CALL_Finish( GET_DISPATCH(), () );
	    __GLX_NOTE_FLUSHED_CMDS(glxc);
	} else {
	    return error;
	}
    }

    pGlxDraw = __glXGetDrawable(glxc, drawId, client, &error);
    if (!pGlxDraw)
	return error;

    if (pGlxDraw == NULL ||
	pGlxDraw->type != GLX_DRAWABLE_WINDOW ||
	pGlxDraw->copySubBuffer == NULL)
	return __glXError(GLXBadDrawable);

    (*pGlxDraw->copySubBuffer)(pGlxDraw, x, y, width, height);

    return Success;
}

/*
** Get drawable attributes
*/
static int
DoGetDrawableAttributes(__GLXclientState *cl, XID drawId)
{
    ClientPtr client = cl->client;
    xGLXGetDrawableAttributesReply reply;
    __GLXdrawable *pGlxDraw;
    CARD32 attributes[6];
    int numAttribs, error;

    if (!validGlxDrawable(client, drawId, GLX_DRAWABLE_ANY,
			  DixGetAttrAccess, &pGlxDraw, &error))
	return error;

    numAttribs = 3;
    reply.length = numAttribs << 1;
    reply.type = X_Reply;
    reply.sequenceNumber = client->sequence;
    reply.numAttribs = numAttribs;

    attributes[0] = GLX_TEXTURE_TARGET_EXT;
    attributes[1] = pGlxDraw->target == GL_TEXTURE_2D ? GLX_TEXTURE_2D_EXT :
	GLX_TEXTURE_RECTANGLE_EXT;
    attributes[2] = GLX_Y_INVERTED_EXT;
    attributes[3] = GL_FALSE;
    attributes[4] = GLX_EVENT_MASK;
    attributes[5] = pGlxDraw->eventMask;

    if (client->swapped) {
	__glXSwapGetDrawableAttributesReply(client, &reply, attributes);
    } else {
	WriteToClient(client, sz_xGLXGetDrawableAttributesReply,
		      (char *)&reply);
	WriteToClient(client, reply.length * sizeof (CARD32),
		      (char *)attributes);
    }

    return Success;
}

int __glXDisp_GetDrawableAttributes(__GLXclientState *cl, GLbyte *pc)
{
    ClientPtr client = cl->client;
    xGLXGetDrawableAttributesReq *req = (xGLXGetDrawableAttributesReq *)pc;

    /* this should be REQUEST_SIZE_MATCH, but mesa sends an additional 4 bytes */
    REQUEST_AT_LEAST_SIZE(xGLXGetDrawableAttributesReq);

    return DoGetDrawableAttributes(cl, req->drawable);
}

int __glXDisp_GetDrawableAttributesSGIX(__GLXclientState *cl, GLbyte *pc)
{
    ClientPtr client = cl->client;
    xGLXGetDrawableAttributesSGIXReq *req =
	(xGLXGetDrawableAttributesSGIXReq *)pc;
    
    REQUEST_SIZE_MATCH(xGLXGetDrawableAttributesSGIXReq);

    return DoGetDrawableAttributes(cl, req->drawable);
}

/************************************************************************/

/*
** Render and Renderlarge are not in the GLX API.  They are used by the GLX
** client library to send batches of GL rendering commands.
*/

/*
** Execute all the drawing commands in a request.
*/
int __glXDisp_Render(__GLXclientState *cl, GLbyte *pc)
{
    xGLXRenderReq *req;
    ClientPtr client= cl->client;
    int left, cmdlen, error;
    int commandsDone;
    CARD16 opcode;
    __GLXrenderHeader *hdr;
    __GLXcontext *glxc;
    __GLX_DECLARE_SWAP_VARIABLES;

    REQUEST_AT_LEAST_SIZE(xGLXRenderReq);

    req = (xGLXRenderReq *) pc;
    if (client->swapped) {
	__GLX_SWAP_SHORT(&req->length);
	__GLX_SWAP_INT(&req->contextTag);
    }

    glxc = __glXForceCurrent(cl, req->contextTag, &error);
    if (!glxc) {
	return error;
    }

    commandsDone = 0;
    pc += sz_xGLXRenderReq;
    left = (req->length << 2) - sz_xGLXRenderReq;
    while (left > 0) {
        __GLXrenderSizeData entry;
        int extra;
	__GLXdispatchRenderProcPtr proc;
	int err;

	if (left < sizeof(__GLXrenderHeader))
	    return BadLength;

	/*
	** Verify that the header length and the overall length agree.
	** Also, each command must be word aligned.
	*/
	hdr = (__GLXrenderHeader *) pc;
	if (client->swapped) {
	    __GLX_SWAP_SHORT(&hdr->length);
	    __GLX_SWAP_SHORT(&hdr->opcode);
	}
	cmdlen = hdr->length;
	opcode = hdr->opcode;

	/*
	** Check for core opcodes and grab entry data.
	*/
	err = __glXGetProtocolSizeData(& Render_dispatch_info, opcode, & entry);
	proc = (__GLXdispatchRenderProcPtr)
	    __glXGetProtocolDecodeFunction(& Render_dispatch_info,
					   opcode, client->swapped);

	if ((err < 0) || (proc == NULL)) {
	    client->errorValue = commandsDone;
	    return __glXError(GLXBadRenderRequest);
	}

        if (entry.varsize) {
            /* variable size command */
            extra = (*entry.varsize)(pc + __GLX_RENDER_HDR_SIZE,
				     client->swapped);
            if (extra < 0) {
                extra = 0;
            }
            if (cmdlen != __GLX_PAD(entry.bytes + extra)) {
                return BadLength;
            }
        } else {
            /* constant size command */
            if (cmdlen != __GLX_PAD(entry.bytes)) {
                return BadLength;
            }
        }
	if (left < cmdlen) {
	    return BadLength;
	}

	/*
	** Skip over the header and execute the command.  We allow the
	** caller to trash the command memory.  This is useful especially
	** for things that require double alignment - they can just shift
	** the data towards lower memory (trashing the header) by 4 bytes
	** and achieve the required alignment.
	*/
	(*proc)(pc + __GLX_RENDER_HDR_SIZE);
	pc += cmdlen;
	left -= cmdlen;
	commandsDone++;
    }
    __GLX_NOTE_UNFLUSHED_CMDS(glxc);
    return Success;
}


/*
** Execute a large rendering request (one that spans multiple X requests).
*/
int __glXDisp_RenderLarge(__GLXclientState *cl, GLbyte *pc)
{
    xGLXRenderLargeReq *req;
    ClientPtr client= cl->client;
    size_t dataBytes;
    __GLXrenderLargeHeader *hdr;
    __GLXcontext *glxc;
    int error;
    CARD16 opcode;
    __GLX_DECLARE_SWAP_VARIABLES;
    
    req = (xGLXRenderLargeReq *) pc;
    if (client->swapped) {
	__GLX_SWAP_SHORT(&req->length);
	__GLX_SWAP_INT(&req->contextTag);
	__GLX_SWAP_INT(&req->dataBytes);
	__GLX_SWAP_SHORT(&req->requestNumber);
	__GLX_SWAP_SHORT(&req->requestTotal);
    }

    glxc = __glXForceCurrent(cl, req->contextTag, &error);
    if (!glxc) {
	/* Reset in case this isn't 1st request. */
	__glXResetLargeCommandStatus(cl);
	return error;
    }
    dataBytes = req->dataBytes;

    /*
    ** Check the request length.
    */
    if ((req->length << 2) != __GLX_PAD(dataBytes) + sz_xGLXRenderLargeReq) {
	client->errorValue = req->length;
	/* Reset in case this isn't 1st request. */
	__glXResetLargeCommandStatus(cl);
	return BadLength;
    }
    pc += sz_xGLXRenderLargeReq;
    
    if (cl->largeCmdRequestsSoFar == 0) {
	__GLXrenderSizeData entry;
	int extra;
	size_t cmdlen;
	int err;

	/*
	** This is the first request of a multi request command.
	** Make enough space in the buffer, then copy the entire request.
	*/
	if (req->requestNumber != 1) {
	    client->errorValue = req->requestNumber;
	    return __glXError(GLXBadLargeRequest);
	}

	hdr = (__GLXrenderLargeHeader *) pc;
	if (client->swapped) {
	    __GLX_SWAP_INT(&hdr->length);
	    __GLX_SWAP_INT(&hdr->opcode);
	}
	cmdlen = hdr->length;
	opcode = hdr->opcode;

	/*
	** Check for core opcodes and grab entry data.
	*/
	err = __glXGetProtocolSizeData(& Render_dispatch_info, opcode, & entry);
	if (err < 0) {
	    client->errorValue = opcode;
	    return __glXError(GLXBadLargeRequest);
	}

	if (entry.varsize) {
	    /*
	    ** If it's a variable-size command (a command whose length must
	    ** be computed from its parameters), all the parameters needed
	    ** will be in the 1st request, so it's okay to do this.
	    */
	    extra = (*entry.varsize)(pc + __GLX_RENDER_LARGE_HDR_SIZE,
				     client->swapped);
	    if (extra < 0) {
		extra = 0;
	    }
	    /* large command's header is 4 bytes longer, so add 4 */
	    if (cmdlen != __GLX_PAD(entry.bytes + 4 + extra)) {
		return BadLength;
	    }
	} else {
	    /* constant size command */
	    if (cmdlen != __GLX_PAD(entry.bytes + 4)) {
		return BadLength;
	    }
	}
	/*
	** Make enough space in the buffer, then copy the entire request.
	*/
	if (cl->largeCmdBufSize < cmdlen) {
	    if (!cl->largeCmdBuf) {
		cl->largeCmdBuf = (GLbyte *) malloc(cmdlen);
	    } else {
		cl->largeCmdBuf = (GLbyte *) realloc(cl->largeCmdBuf, cmdlen);
	    }
	    if (!cl->largeCmdBuf) {
		return BadAlloc;
	    }
	    cl->largeCmdBufSize = cmdlen;
	}
	memcpy(cl->largeCmdBuf, pc, dataBytes);

	cl->largeCmdBytesSoFar = dataBytes;
	cl->largeCmdBytesTotal = cmdlen;
	cl->largeCmdRequestsSoFar = 1;
	cl->largeCmdRequestsTotal = req->requestTotal;
	return Success;
	
    } else {
	/*
	** We are receiving subsequent (i.e. not the first) requests of a
	** multi request command.
	*/

	/*
	** Check the request number and the total request count.
	*/
	if (req->requestNumber != cl->largeCmdRequestsSoFar + 1) {
	    client->errorValue = req->requestNumber;
	    __glXResetLargeCommandStatus(cl);
	    return __glXError(GLXBadLargeRequest);
	}
	if (req->requestTotal != cl->largeCmdRequestsTotal) {
	    client->errorValue = req->requestTotal;
	    __glXResetLargeCommandStatus(cl);
	    return __glXError(GLXBadLargeRequest);
	}

	/*
	** Check that we didn't get too much data.
	*/
	if ((cl->largeCmdBytesSoFar + dataBytes) > cl->largeCmdBytesTotal) {
	    client->errorValue = dataBytes;
	    __glXResetLargeCommandStatus(cl);
	    return __glXError(GLXBadLargeRequest);
	}
	memcpy(cl->largeCmdBuf + cl->largeCmdBytesSoFar, pc, dataBytes);
	cl->largeCmdBytesSoFar += dataBytes;
	cl->largeCmdRequestsSoFar++;

	if (req->requestNumber == cl->largeCmdRequestsTotal) {
	    __GLXdispatchRenderProcPtr proc;

	    /*
	    ** This is the last request; it must have enough bytes to complete
	    ** the command.
	    */
	    /* NOTE: the two pad macros have been added below; they are needed
	    ** because the client library pads the total byte count, but not
	    ** the per-request byte counts.  The Protocol Encoding says the
	    ** total byte count should not be padded, so a proposal will be 
	    ** made to the ARB to relax the padding constraint on the total 
	    ** byte count, thus preserving backward compatibility.  Meanwhile, 
	    ** the padding done below fixes a bug that did not allow
	    ** large commands of odd sizes to be accepted by the server.
	    */
	    if (__GLX_PAD(cl->largeCmdBytesSoFar) !=
		__GLX_PAD(cl->largeCmdBytesTotal)) {
		client->errorValue = dataBytes;
		__glXResetLargeCommandStatus(cl);
		return __glXError(GLXBadLargeRequest);
	    }
	    hdr = (__GLXrenderLargeHeader *) cl->largeCmdBuf;
	    /*
	    ** The opcode and length field in the header had already been
	    ** swapped when the first request was received.
	    **
	    ** Use the opcode to index into the procedure table.
	    */
	    opcode = hdr->opcode;

	    proc = (__GLXdispatchRenderProcPtr)
	      __glXGetProtocolDecodeFunction(& Render_dispatch_info, opcode,
					     client->swapped);
	    if (proc == NULL) {
		client->errorValue = opcode;
		return __glXError(GLXBadLargeRequest);
	    }

	    /*
	    ** Skip over the header and execute the command.
	    */
	    (*proc)(cl->largeCmdBuf + __GLX_RENDER_LARGE_HDR_SIZE);
	    __GLX_NOTE_UNFLUSHED_CMDS(glxc);

	    /*
	    ** Reset for the next RenderLarge series.
	    */
	    __glXResetLargeCommandStatus(cl);
	} else {
	    /*
	    ** This is neither the first nor the last request.
	    */
	}
	return Success;
    }
}

/************************************************************************/

/*
** No support is provided for the vendor-private requests other than
** allocating the entry points in the dispatch table.
*/

int __glXDisp_VendorPrivate(__GLXclientState *cl, GLbyte *pc)
{
    ClientPtr client = cl->client;
    xGLXVendorPrivateReq *req = (xGLXVendorPrivateReq *) pc;
    GLint vendorcode = req->vendorCode;
    __GLXdispatchVendorPrivProcPtr proc;

    REQUEST_AT_LEAST_SIZE(xGLXVendorPrivateReq);

    proc = (__GLXdispatchVendorPrivProcPtr)
      __glXGetProtocolDecodeFunction(& VendorPriv_dispatch_info,
				     vendorcode, 0);
    if (proc != NULL) {
	(*proc)(cl, (GLbyte*)req);
	return Success;
    }

    cl->client->errorValue = req->vendorCode;
    return __glXError(GLXUnsupportedPrivateRequest);
}

int __glXDisp_VendorPrivateWithReply(__GLXclientState *cl, GLbyte *pc)
{
    ClientPtr client = cl->client;
    xGLXVendorPrivateReq *req = (xGLXVendorPrivateReq *) pc;
    GLint vendorcode = req->vendorCode;
    __GLXdispatchVendorPrivProcPtr proc;

    REQUEST_AT_LEAST_SIZE(xGLXVendorPrivateReq);

    proc = (__GLXdispatchVendorPrivProcPtr)
      __glXGetProtocolDecodeFunction(& VendorPriv_dispatch_info,
				     vendorcode, 0);
    if (proc != NULL) {
	return (*proc)(cl, (GLbyte*)req);
    }

    cl->client->errorValue = vendorcode;
    return __glXError(GLXUnsupportedPrivateRequest);
}

int __glXDisp_QueryExtensionsString(__GLXclientState *cl, GLbyte *pc)
{
    ClientPtr client = cl->client;
    xGLXQueryExtensionsStringReq *req = (xGLXQueryExtensionsStringReq *) pc;
    xGLXQueryExtensionsStringReply reply;
    __GLXscreen *pGlxScreen;
    size_t n, length;
    char *buf;
    int err;

    REQUEST_SIZE_MATCH(xGLXQueryExtensionsStringReq);

    if (!validGlxScreen(client, req->screen, &pGlxScreen, &err))
	return err;

    n = strlen(pGlxScreen->GLXextensions) + 1;
    length = __GLX_PAD(n) >> 2;
    reply.type = X_Reply;
    reply.sequenceNumber = client->sequence;
    reply.length = length;
    reply.n = n;

    /* Allocate buffer to make sure it's a multiple of 4 bytes big.*/
    buf = (char *) malloc(length << 2);
    if (buf == NULL)
        return BadAlloc;
    memcpy(buf, pGlxScreen->GLXextensions, n);

    if (client->swapped) {
        glxSwapQueryExtensionsStringReply(client, &reply, buf);
    } else {
        WriteToClient(client, sz_xGLXQueryExtensionsStringReply,(char *)&reply);
        WriteToClient(client, (int)(length << 2), (char *)buf);
    }

    free(buf);
    return Success;
}

int __glXDisp_QueryServerString(__GLXclientState *cl, GLbyte *pc)
{
    ClientPtr client = cl->client;
    xGLXQueryServerStringReq *req = (xGLXQueryServerStringReq *) pc;
    xGLXQueryServerStringReply reply;
    size_t n, length;
    const char *ptr;
    char *buf;
    __GLXscreen *pGlxScreen;
    int err;
    char ver_str[16];

    REQUEST_SIZE_MATCH(xGLXQueryServerStringReq);

    if (!validGlxScreen(client, req->screen, &pGlxScreen, &err))
	return err;

    switch(req->name) {
	case GLX_VENDOR:
	    ptr = pGlxScreen->GLXvendor;
	    break;
	case GLX_VERSION:
	    /* Return to the server version rather than the screen version
	     * to prevent confusion when they do not match.
	     */
	    snprintf(ver_str, 16, "%d.%d", glxMajorVersion, glxMinorVersion);
	    ptr = ver_str;
	    break;
	case GLX_EXTENSIONS:
	    ptr = pGlxScreen->GLXextensions;
	    break;
	default:
	    return BadValue; 
    }

    n = strlen(ptr) + 1;
    length = __GLX_PAD(n) >> 2;
    reply.type = X_Reply;
    reply.sequenceNumber = client->sequence;
    reply.length = length;
    reply.n = n;

    buf = (char *) malloc(length << 2);
    if (buf == NULL) {
        return BadAlloc;
    }
    memcpy(buf, ptr, n);

    if (client->swapped) {
        glxSwapQueryServerStringReply(client, &reply, buf);
    } else {
        WriteToClient(client, sz_xGLXQueryServerStringReply, (char *)&reply);
        WriteToClient(client, (int)(length << 2), buf);
    }

    free(buf);
    return Success;
}

int __glXDisp_ClientInfo(__GLXclientState *cl, GLbyte *pc)
{
    ClientPtr client = cl->client;
    xGLXClientInfoReq *req = (xGLXClientInfoReq *) pc;
    const char *buf;
   
    REQUEST_AT_LEAST_SIZE(xGLXClientInfoReq);

    buf = (const char *)(req+1);
    if (!memchr(buf, 0, (client->req_len << 2) - sizeof(xGLXClientInfoReq)))
	return BadLength;

    cl->GLClientmajorVersion = req->major;
    cl->GLClientminorVersion = req->minor;
    free(cl->GLClientextensions);
    cl->GLClientextensions = strdup(buf);

    return Success;
}