aboutsummaryrefslogtreecommitdiff
path: root/nx-X11/extras/Mesa/src/mesa/drivers/dri/unichrome/via_tex.c
diff options
context:
space:
mode:
Diffstat (limited to 'nx-X11/extras/Mesa/src/mesa/drivers/dri/unichrome/via_tex.c')
-rw-r--r--nx-X11/extras/Mesa/src/mesa/drivers/dri/unichrome/via_tex.c970
1 files changed, 0 insertions, 970 deletions
diff --git a/nx-X11/extras/Mesa/src/mesa/drivers/dri/unichrome/via_tex.c b/nx-X11/extras/Mesa/src/mesa/drivers/dri/unichrome/via_tex.c
deleted file mode 100644
index c308339f2..000000000
--- a/nx-X11/extras/Mesa/src/mesa/drivers/dri/unichrome/via_tex.c
+++ /dev/null
@@ -1,970 +0,0 @@
-/*
- * Copyright 1998-2003 VIA Technologies, Inc. All Rights Reserved.
- * Copyright 2001-2003 S3 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, 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
- * VIA, S3 GRAPHICS, AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR
- * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
- * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
- * DEALINGS IN THE SOFTWARE.
- */
-
-
-#include <stdlib.h>
-#include <stdio.h>
-
-#include "glheader.h"
-#include "macros.h"
-#include "mtypes.h"
-#include "enums.h"
-#include "colortab.h"
-#include "convolve.h"
-#include "context.h"
-#include "simple_list.h"
-#include "texcompress.h"
-#include "texformat.h"
-#include "texobj.h"
-#include "texstore.h"
-
-#include "mm.h"
-#include "via_context.h"
-#include "via_fb.h"
-#include "via_tex.h"
-#include "via_state.h"
-#include "via_ioctl.h"
-#include "via_3d_reg.h"
-
-static const struct gl_texture_format *
-viaChooseTexFormat( GLcontext *ctx, GLint internalFormat,
- GLenum format, GLenum type )
-{
- struct via_context *vmesa = VIA_CONTEXT(ctx);
- const GLboolean do32bpt = ( vmesa->viaScreen->bitsPerPixel == 32
-/* && vmesa->viaScreen->textureSize > 4*1024*1024 */
- );
-
-
- switch ( internalFormat ) {
- case 4:
- case GL_RGBA:
- case GL_COMPRESSED_RGBA:
- if ( format == GL_BGRA ) {
- if ( type == GL_UNSIGNED_INT_8_8_8_8_REV ||
- type == GL_UNSIGNED_BYTE ) {
- return &_mesa_texformat_argb8888;
- }
- else if ( type == GL_UNSIGNED_SHORT_4_4_4_4_REV ) {
- return &_mesa_texformat_argb4444;
- }
- else if ( type == GL_UNSIGNED_SHORT_1_5_5_5_REV ) {
- return &_mesa_texformat_argb1555;
- }
- }
- else if ( type == GL_UNSIGNED_BYTE ||
- type == GL_UNSIGNED_INT_8_8_8_8_REV ||
- type == GL_UNSIGNED_INT_8_8_8_8 ) {
- return &_mesa_texformat_argb8888;
- }
- return do32bpt ? &_mesa_texformat_argb8888 : &_mesa_texformat_argb4444;
-
- case 3:
- case GL_RGB:
- case GL_COMPRESSED_RGB:
- if ( format == GL_RGB && type == GL_UNSIGNED_SHORT_5_6_5 ) {
- return &_mesa_texformat_rgb565;
- }
- else if ( type == GL_UNSIGNED_BYTE ) {
- return &_mesa_texformat_argb8888;
- }
- return do32bpt ? &_mesa_texformat_argb8888 : &_mesa_texformat_rgb565;
-
- case GL_RGBA8:
- case GL_RGB10_A2:
- case GL_RGBA12:
- case GL_RGBA16:
- return &_mesa_texformat_argb8888;
-
- case GL_RGBA4:
- case GL_RGBA2:
- return &_mesa_texformat_argb4444;
-
- case GL_RGB5_A1:
- return &_mesa_texformat_argb1555;
-
- case GL_RGB8:
- case GL_RGB10:
- case GL_RGB12:
- case GL_RGB16:
- return &_mesa_texformat_argb8888;
-
- case GL_RGB5:
- case GL_RGB4:
- case GL_R3_G3_B2:
- return &_mesa_texformat_rgb565;
-
- case GL_ALPHA:
- case GL_ALPHA4:
- case GL_ALPHA8:
- case GL_ALPHA12:
- case GL_ALPHA16:
- case GL_COMPRESSED_ALPHA:
- return &_mesa_texformat_a8;
-
- case 1:
- case GL_LUMINANCE:
- case GL_LUMINANCE4:
- case GL_LUMINANCE8:
- case GL_LUMINANCE12:
- case GL_LUMINANCE16:
- case GL_COMPRESSED_LUMINANCE:
- return &_mesa_texformat_l8;
-
- case 2:
- case GL_LUMINANCE_ALPHA:
- case GL_LUMINANCE4_ALPHA4:
- case GL_LUMINANCE6_ALPHA2:
- case GL_LUMINANCE8_ALPHA8:
- case GL_LUMINANCE12_ALPHA4:
- case GL_LUMINANCE12_ALPHA12:
- case GL_LUMINANCE16_ALPHA16:
- case GL_COMPRESSED_LUMINANCE_ALPHA:
- return &_mesa_texformat_al88;
-
- case GL_INTENSITY:
- case GL_INTENSITY4:
- case GL_INTENSITY8:
- case GL_INTENSITY12:
- case GL_INTENSITY16:
- case GL_COMPRESSED_INTENSITY:
- return &_mesa_texformat_i8;
-
- case GL_YCBCR_MESA:
- if (type == GL_UNSIGNED_SHORT_8_8_MESA ||
- type == GL_UNSIGNED_BYTE)
- return &_mesa_texformat_ycbcr;
- else
- return &_mesa_texformat_ycbcr_rev;
-
- case GL_COMPRESSED_RGB_FXT1_3DFX:
- return &_mesa_texformat_rgb_fxt1;
- case GL_COMPRESSED_RGBA_FXT1_3DFX:
- return &_mesa_texformat_rgba_fxt1;
-
- case GL_RGB_S3TC:
- case GL_RGB4_S3TC:
- case GL_COMPRESSED_RGB_S3TC_DXT1_EXT:
- return &_mesa_texformat_rgb_dxt1;
-
- case GL_COMPRESSED_RGBA_S3TC_DXT1_EXT:
- return &_mesa_texformat_rgba_dxt1;
-
- case GL_RGBA_S3TC:
- case GL_RGBA4_S3TC:
- case GL_COMPRESSED_RGBA_S3TC_DXT3_EXT:
- return &_mesa_texformat_rgba_dxt3;
-
- case GL_COMPRESSED_RGBA_S3TC_DXT5_EXT:
- return &_mesa_texformat_rgba_dxt5;
-
- case GL_COLOR_INDEX:
- case GL_COLOR_INDEX1_EXT:
- case GL_COLOR_INDEX2_EXT:
- case GL_COLOR_INDEX4_EXT:
- case GL_COLOR_INDEX8_EXT:
- case GL_COLOR_INDEX12_EXT:
- case GL_COLOR_INDEX16_EXT:
- return &_mesa_texformat_ci8;
-
- default:
- fprintf(stderr, "unexpected texture format %s in %s\n",
- _mesa_lookup_enum_by_nr(internalFormat),
- __FUNCTION__);
- return NULL;
- }
-
- return NULL; /* never get here */
-}
-
-static int logbase2(int n)
-{
- GLint i = 1;
- GLint log2 = 0;
-
- while (n > i) {
- i *= 2;
- log2++;
- }
-
- return log2;
-}
-
-static const char *get_memtype_name( GLint memType )
-{
- static const char *names[] = {
- "VIA_MEM_VIDEO",
- "VIA_MEM_AGP",
- "VIA_MEM_SYSTEM",
- "VIA_MEM_MIXED",
- "VIA_MEM_UNKNOWN"
- };
-
- return names[memType];
-}
-
-
-static GLboolean viaMoveTexBuffers( struct via_context *vmesa,
- struct via_tex_buffer **buffers,
- GLuint nr,
- GLint newMemType )
-{
- struct via_tex_buffer *newTexBuf[VIA_MAX_TEXLEVELS];
- GLint i;
-
- if (VIA_DEBUG & DEBUG_TEXTURE)
- fprintf(stderr, "%s to %s\n",
- __FUNCTION__,
- get_memtype_name(newMemType));
-
- memset(newTexBuf, 0, sizeof(newTexBuf));
-
- /* First do all the allocations (or fail):
- */
- for (i = 0; i < nr; i++) {
- if (buffers[i]->memType != newMemType) {
-
- /* Don't allow uploads in a thrash state. Should try and
- * catch this earlier.
- */
- if (vmesa->thrashing && newMemType != VIA_MEM_SYSTEM)
- goto cleanup;
-
- newTexBuf[i] = via_alloc_texture(vmesa,
- buffers[i]->size,
- newMemType);
- if (!newTexBuf[i])
- goto cleanup;
- }
- }
-
-
- /* Now copy all the image data and free the old texture memory.
- */
- for (i = 0; i < nr; i++) {
- if (newTexBuf[i]) {
- memcpy(newTexBuf[i]->bufAddr,
- buffers[i]->bufAddr,
- buffers[i]->size);
-
- newTexBuf[i]->image = buffers[i]->image;
- newTexBuf[i]->image->texMem = newTexBuf[i];
- newTexBuf[i]->image->image.Data = newTexBuf[i]->bufAddr;
- via_free_texture(vmesa, buffers[i]);
- }
- }
-
- if (VIA_DEBUG & DEBUG_TEXTURE)
- fprintf(stderr, "%s - success\n", __FUNCTION__);
-
- return GL_TRUE;
-
- cleanup:
- /* Release any allocations made prior to failure:
- */
- if (VIA_DEBUG & DEBUG_TEXTURE)
- fprintf(stderr, "%s - failed\n", __FUNCTION__);
-
- for (i = 0; i < nr; i++) {
- if (newTexBuf[i]) {
- via_free_texture(vmesa, newTexBuf[i]);
- }
- }
-
- return GL_FALSE;
-}
-
-
-static GLboolean viaMoveTexObject( struct via_context *vmesa,
- struct via_texture_object *viaObj,
- GLint newMemType )
-{
- struct via_texture_image **viaImage =
- (struct via_texture_image **)&viaObj->obj.Image[0][0];
- struct via_tex_buffer *buffers[VIA_MAX_TEXLEVELS];
- GLuint i, nr = 0;
-
- for (i = viaObj->firstLevel; i <= viaObj->lastLevel; i++)
- buffers[nr++] = viaImage[i]->texMem;
-
- if (viaMoveTexBuffers( vmesa, &buffers[0], nr, newMemType )) {
- viaObj->memType = newMemType;
- return GL_TRUE;
- }
-
- return GL_FALSE;
-}
-
-
-
-static GLboolean viaSwapInTexObject( struct via_context *vmesa,
- struct via_texture_object *viaObj )
-{
- const struct via_texture_image *baseImage =
- (struct via_texture_image *)viaObj->obj.Image[0][viaObj->obj.BaseLevel];
-
- if (VIA_DEBUG & DEBUG_TEXTURE)
- fprintf(stderr, "%s\n", __FUNCTION__);
-
- if (baseImage->texMem->memType != VIA_MEM_SYSTEM)
- return viaMoveTexObject( vmesa, viaObj, baseImage->texMem->memType );
-
- return (viaMoveTexObject( vmesa, viaObj, VIA_MEM_AGP ) ||
- viaMoveTexObject( vmesa, viaObj, VIA_MEM_VIDEO ));
-}
-
-
-/* This seems crude, but it asks a fairly pertinent question and gives
- * an accurate answer:
- */
-static GLboolean viaIsTexMemLow( struct via_context *vmesa,
- GLuint heap )
-{
- struct via_tex_buffer *buf = via_alloc_texture(vmesa, 512 * 1024, heap );
- if (!buf)
- return GL_TRUE;
-
- via_free_texture(vmesa, buf);
- return GL_FALSE;
-}
-
-
-/* Speculatively move texture images which haven't been used in a
- * while back to system memory.
- *
- * TODO: only do this when texture memory is low.
- *
- * TODO: use dma.
- *
- * TODO: keep the fb/agp version hanging around and use the local
- * version as backing store, so re-upload might be avoided.
- *
- * TODO: do this properly in the kernel...
- */
-GLboolean viaSwapOutWork( struct via_context *vmesa )
-{
- struct via_tex_buffer *s, *tmp;
- GLuint done = 0;
- GLuint heap, target;
-
- if (VIA_DEBUG & DEBUG_TEXTURE)
- fprintf(stderr, "%s VID %d AGP %d SYS %d\n", __FUNCTION__,
- vmesa->total_alloc[VIA_MEM_VIDEO],
- vmesa->total_alloc[VIA_MEM_AGP],
- vmesa->total_alloc[VIA_MEM_SYSTEM]);
-
-
- for (heap = VIA_MEM_VIDEO; heap <= VIA_MEM_AGP; heap++) {
- GLuint nr = 0, sz = 0;
-
- if (vmesa->thrashing) {
- if (VIA_DEBUG & DEBUG_TEXTURE)
- fprintf(stderr, "Heap %d: trash flag\n", heap);
- target = 1*1024*1024;
- }
- else if (viaIsTexMemLow(vmesa, heap)) {
- if (VIA_DEBUG & DEBUG_TEXTURE)
- fprintf(stderr, "Heap %d: low memory\n", heap);
- target = 64*1024;
- }
- else {
- if (VIA_DEBUG & DEBUG_TEXTURE)
- fprintf(stderr, "Heap %d: nothing to do\n", heap);
- continue;
- }
-
- foreach_s( s, tmp, &vmesa->tex_image_list[heap] ) {
- if (s->lastUsed < vmesa->lastSwap[1]) {
- struct via_texture_object *viaObj =
- (struct via_texture_object *) s->image->image.TexObject;
-
- if (VIA_DEBUG & DEBUG_TEXTURE)
- fprintf(stderr,
- "back copy tex sz %d, lastUsed %d lastSwap %d\n",
- s->size, s->lastUsed, vmesa->lastSwap[1]);
-
- if (viaMoveTexBuffers( vmesa, &s, 1, VIA_MEM_SYSTEM )) {
- viaObj->memType = VIA_MEM_MIXED;
- done += s->size;
- }
- else {
- if (VIA_DEBUG & DEBUG_TEXTURE)
- fprintf(stderr, "Failed to back copy texture!\n");
- sz += s->size;
- }
- }
- else {
- nr ++;
- sz += s->size;
- }
-
- if (done > target) {
- vmesa->thrashing = GL_FALSE; /* might not get set otherwise? */
- return GL_TRUE;
- }
- }
-
- assert(sz == vmesa->total_alloc[heap]);
-
- if (VIA_DEBUG & DEBUG_TEXTURE)
- fprintf(stderr, "Heap %d: nr %d tot sz %d\n", heap, nr, sz);
- }
-
-
- return done != 0;
-}
-
-
-
-/* Basically, just collect the image dimensions and addresses for each
- * image and update the texture object state accordingly.
- */
-static GLboolean viaSetTexImages(GLcontext *ctx,
- struct gl_texture_object *texObj)
-{
- struct via_context *vmesa = VIA_CONTEXT(ctx);
- struct via_texture_object *viaObj = (struct via_texture_object *)texObj;
- const struct via_texture_image *baseImage =
- (struct via_texture_image *)texObj->Image[0][texObj->BaseLevel];
- GLint firstLevel, lastLevel, numLevels;
- GLuint texFormat;
- GLint w, h, p;
- GLint i, j = 0, k = 0, l = 0, m = 0;
- GLuint texBase;
- GLuint basH = 0;
- GLuint widthExp = 0;
- GLuint heightExp = 0;
-
- switch (baseImage->image.TexFormat->MesaFormat) {
- case MESA_FORMAT_ARGB8888:
- texFormat = HC_HTXnFM_ARGB8888;
- break;
- case MESA_FORMAT_ARGB4444:
- texFormat = HC_HTXnFM_ARGB4444;
- break;
- case MESA_FORMAT_RGB565:
- texFormat = HC_HTXnFM_RGB565;
- break;
- case MESA_FORMAT_ARGB1555:
- texFormat = HC_HTXnFM_ARGB1555;
- break;
- case MESA_FORMAT_RGB888:
- texFormat = HC_HTXnFM_ARGB0888;
- break;
- case MESA_FORMAT_L8:
- texFormat = HC_HTXnFM_L8;
- break;
- case MESA_FORMAT_I8:
- texFormat = HC_HTXnFM_T8;
- break;
- case MESA_FORMAT_CI8:
- texFormat = HC_HTXnFM_Index8;
- break;
- case MESA_FORMAT_AL88:
- texFormat = HC_HTXnFM_AL88;
- break;
- case MESA_FORMAT_A8:
- texFormat = HC_HTXnFM_A8;
- break;
- default:
- _mesa_problem(vmesa->glCtx, "Bad texture format in viaSetTexImages");
- return GL_FALSE;
- }
-
- /* Compute which mipmap levels we really want to send to the hardware.
- * This depends on the base image size, GL_TEXTURE_MIN_LOD,
- * GL_TEXTURE_MAX_LOD, GL_TEXTURE_BASE_LEVEL, and GL_TEXTURE_MAX_LEVEL.
- * Yes, this looks overly complicated, but it's all needed.
- */
- if (texObj->MinFilter == GL_LINEAR || texObj->MinFilter == GL_NEAREST) {
- firstLevel = lastLevel = texObj->BaseLevel;
- }
- else {
- firstLevel = texObj->BaseLevel + (GLint)(texObj->MinLod + 0.5);
- firstLevel = MAX2(firstLevel, texObj->BaseLevel);
- lastLevel = texObj->BaseLevel + (GLint)(texObj->MaxLod + 0.5);
- lastLevel = MAX2(lastLevel, texObj->BaseLevel);
- lastLevel = MIN2(lastLevel, texObj->BaseLevel + baseImage->image.MaxLog2);
- lastLevel = MIN2(lastLevel, texObj->MaxLevel);
- lastLevel = MAX2(firstLevel, lastLevel); /* need at least one level */
- }
-
- numLevels = lastLevel - firstLevel + 1;
-
- /* save these values, check if they effect the residency of the
- * texture:
- */
- if (viaObj->firstLevel != firstLevel ||
- viaObj->lastLevel != lastLevel) {
- viaObj->firstLevel = firstLevel;
- viaObj->lastLevel = lastLevel;
- viaObj->memType = VIA_MEM_MIXED;
- }
-
- if (VIA_DEBUG & DEBUG_TEXTURE & 0)
- fprintf(stderr, "%s, current memType: %s\n",
- __FUNCTION__,
- get_memtype_name(viaObj->memType));
-
-
- if (viaObj->memType == VIA_MEM_MIXED ||
- viaObj->memType == VIA_MEM_SYSTEM) {
- if (!viaSwapInTexObject(vmesa, viaObj)) {
- if (VIA_DEBUG & DEBUG_TEXTURE)
- if (!vmesa->thrashing)
- fprintf(stderr, "Thrashing flag set for frame %d\n",
- vmesa->swap_count);
- vmesa->thrashing = GL_TRUE;
- return GL_FALSE;
- }
- }
-
- if (viaObj->memType == VIA_MEM_AGP)
- viaObj->regTexFM = (HC_SubA_HTXnFM << 24) | HC_HTXnLoc_AGP | texFormat;
- else
- viaObj->regTexFM = (HC_SubA_HTXnFM << 24) | HC_HTXnLoc_Local | texFormat;
-
-
- for (i = 0; i < numLevels; i++) {
- struct via_texture_image *viaImage =
- (struct via_texture_image *)texObj->Image[0][firstLevel + i];
-
- w = viaImage->image.WidthLog2;
- h = viaImage->image.HeightLog2;
- p = viaImage->pitchLog2;
-
- assert(viaImage->texMem->memType == viaObj->memType);
-
- texBase = viaImage->texMem->texBase;
- if (!texBase) {
- if (VIA_DEBUG & DEBUG_TEXTURE)
- fprintf(stderr, "%s: no texBase[%d]\n", __FUNCTION__, i);
- return GL_FALSE;
- }
-
- /* Image has to remain resident until the coming fence is retired.
- */
- move_to_head( &vmesa->tex_image_list[viaImage->texMem->memType],
- viaImage->texMem );
- viaImage->texMem->lastUsed = vmesa->lastBreadcrumbWrite;
-
-
- viaObj->regTexBaseAndPitch[i].baseL =
- ((HC_SubA_HTXnL0BasL + i) << 24) | (texBase & 0xFFFFFF);
-
- viaObj->regTexBaseAndPitch[i].pitchLog2 =
- ((HC_SubA_HTXnL0Pit + i) << 24) | (p << 20);
-
-
- /* The base high bytes for each 3 levels are packed
- * together into one register:
- */
- j = i / 3;
- k = 3 - (i % 3);
- basH |= ((texBase & 0xFF000000) >> (k << 3));
- if (k == 1) {
- viaObj->regTexBaseH[j] = ((j + HC_SubA_HTXnL012BasH) << 24) | basH;
- basH = 0;
- }
-
- /* Likewise, sets of 6 log2width and log2height values are
- * packed into individual registers:
- */
- l = i / 6;
- m = i % 6;
- widthExp |= (((GLuint)w & 0xF) << (m << 2));
- heightExp |= (((GLuint)h & 0xF) << (m << 2));
- if (m == 5) {
- viaObj->regTexWidthLog2[l] =
- (l + HC_SubA_HTXnL0_5WE) << 24 | widthExp;
- viaObj->regTexHeightLog2[l] =
- (l + HC_SubA_HTXnL0_5HE) << 24 | heightExp;
- widthExp = 0;
- heightExp = 0;
- }
- if (w) w--;
- if (h) h--;
- if (p) p--;
- }
-
- if (k != 1) {
- viaObj->regTexBaseH[j] = ((j + HC_SubA_HTXnL012BasH) << 24) | basH;
- }
- if (m != 5) {
- viaObj->regTexWidthLog2[l] = (l + HC_SubA_HTXnL0_5WE) << 24 | widthExp;
- viaObj->regTexHeightLog2[l] = (l + HC_SubA_HTXnL0_5HE) << 24 | heightExp;
- }
-
- return GL_TRUE;
-}
-
-
-GLboolean viaUpdateTextureState( GLcontext *ctx )
-{
- struct gl_texture_unit *texUnit = ctx->Texture.Unit;
- GLuint i;
-
- for (i = 0; i < 2; i++) {
- if (texUnit[i]._ReallyEnabled == TEXTURE_2D_BIT ||
- texUnit[i]._ReallyEnabled == TEXTURE_1D_BIT) {
-
- if (!viaSetTexImages(ctx, texUnit[i]._Current))
- return GL_FALSE;
- }
- else if (texUnit[i]._ReallyEnabled) {
- return GL_FALSE;
- }
- }
-
- return GL_TRUE;
-}
-
-
-
-
-
-
-
-
-static void viaTexImage(GLcontext *ctx,
- GLint dims,
- GLenum target, GLint level,
- GLint internalFormat,
- GLint width, GLint height, GLint border,
- GLenum format, GLenum type, const void *pixels,
- const struct gl_pixelstore_attrib *packing,
- struct gl_texture_object *texObj,
- struct gl_texture_image *texImage)
-{
- struct via_context *vmesa = VIA_CONTEXT(ctx);
- GLint postConvWidth = width;
- GLint postConvHeight = height;
- GLint texelBytes, sizeInBytes;
- struct via_texture_object *viaObj = (struct via_texture_object *)texObj;
- struct via_texture_image *viaImage = (struct via_texture_image *)texImage;
- int heaps[3], nheaps, i;
-
- if (!is_empty_list(&vmesa->freed_tex_buffers)) {
- viaCheckBreadcrumb(vmesa, 0);
- via_release_pending_textures(vmesa);
- }
-
- if (ctx->_ImageTransferState & IMAGE_CONVOLUTION_BIT) {
- _mesa_adjust_image_for_convolution(ctx, dims, &postConvWidth,
- &postConvHeight);
- }
-
- /* choose the texture format */
- texImage->TexFormat = viaChooseTexFormat(ctx, internalFormat,
- format, type);
-
- assert(texImage->TexFormat);
-
- if (dims == 1) {
- texImage->FetchTexelc = texImage->TexFormat->FetchTexel1D;
- texImage->FetchTexelf = texImage->TexFormat->FetchTexel1Df;
- }
- else {
- texImage->FetchTexelc = texImage->TexFormat->FetchTexel2D;
- texImage->FetchTexelf = texImage->TexFormat->FetchTexel2Df;
- }
- texelBytes = texImage->TexFormat->TexelBytes;
-
-
- /* Minimum pitch of 32 bytes */
- if (postConvWidth * texelBytes < 32) {
- postConvWidth = 32 / texelBytes;
- texImage->RowStride = postConvWidth;
- }
-
- assert(texImage->RowStride == postConvWidth);
- viaImage->pitchLog2 = logbase2(postConvWidth * texelBytes);
-
- /* allocate memory */
- if (texImage->IsCompressed)
- sizeInBytes = texImage->CompressedSize;
- else
- sizeInBytes = postConvWidth * postConvHeight * texelBytes;
-
-
- /* Attempt to allocate texture memory directly, otherwise use main
- * memory and this texture will always be a fallback. FIXME!
- *
- * TODO: make room in agp if this fails.
- * TODO: use fb ram for textures as well.
- */
-
-
- switch (viaObj->memType) {
- case VIA_MEM_UNKNOWN:
- heaps[0] = VIA_MEM_AGP;
- heaps[1] = VIA_MEM_VIDEO;
- heaps[2] = VIA_MEM_SYSTEM;
- nheaps = 3;
- break;
- case VIA_MEM_AGP:
- case VIA_MEM_VIDEO:
- heaps[0] = viaObj->memType;
- heaps[1] = VIA_MEM_SYSTEM;
- nheaps = 2;
- break;
- case VIA_MEM_MIXED:
- case VIA_MEM_SYSTEM:
- default:
- heaps[0] = VIA_MEM_SYSTEM;
- nheaps = 1;
- break;
- }
-
- for (i = 0; i < nheaps && !viaImage->texMem; i++) {
- if (VIA_DEBUG & DEBUG_TEXTURE)
- fprintf(stderr, "try %s (obj %s)\n", get_memtype_name(heaps[i]),
- get_memtype_name(viaObj->memType));
- viaImage->texMem = via_alloc_texture(vmesa, sizeInBytes, heaps[i]);
- }
-
- if (!viaImage->texMem) {
- _mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexImage");
- return;
- }
-
- if (VIA_DEBUG & DEBUG_TEXTURE)
- fprintf(stderr, "upload %d bytes to %s\n", sizeInBytes,
- get_memtype_name(viaImage->texMem->memType));
-
- viaImage->texMem->image = viaImage;
- texImage->Data = viaImage->texMem->bufAddr;
-
- if (viaObj->memType == VIA_MEM_UNKNOWN)
- viaObj->memType = viaImage->texMem->memType;
- else if (viaObj->memType != viaImage->texMem->memType)
- viaObj->memType = VIA_MEM_MIXED;
-
- if (VIA_DEBUG & DEBUG_TEXTURE)
- fprintf(stderr, "%s, obj %s, image : %s\n",
- __FUNCTION__,
- get_memtype_name(viaObj->memType),
- get_memtype_name(viaImage->texMem->memType));
-
- vmesa->clearTexCache = 1;
-
- pixels = _mesa_validate_pbo_teximage(ctx, dims, width, height, 1,
- format, type,
- pixels, packing, "glTexImage");
- if (!pixels) {
- /* Note: we check for a NULL image pointer here, _after_ we allocated
- * memory for the texture. That's what the GL spec calls for.
- */
- return;
- }
- else {
- GLint dstRowStride, dstImageStride = 0;
- GLboolean success;
- if (texImage->IsCompressed) {
- dstRowStride = _mesa_compressed_row_stride(texImage->IntFormat,width);
- }
- else {
- dstRowStride = postConvWidth * texImage->TexFormat->TexelBytes;
- }
- ASSERT(texImage->TexFormat->StoreImage);
- success = texImage->TexFormat->StoreImage(ctx, dims, texImage->Format,
- texImage->TexFormat,
- texImage->Data,
- 0, 0, 0, /* dstX/Y/Zoffset */
- dstRowStride, dstImageStride,
- width, height, 1,
- format, type, pixels, packing);
- if (!success) {
- _mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexImage");
- }
- }
-
- /* GL_SGIS_generate_mipmap */
- if (level == texObj->BaseLevel && texObj->GenerateMipmap) {
- _mesa_generate_mipmap(ctx, target,
- &ctx->Texture.Unit[ctx->Texture.CurrentUnit],
- texObj);
- }
-
- _mesa_unmap_teximage_pbo(ctx, packing);
-}
-
-static void viaTexImage2D(GLcontext *ctx,
- GLenum target, GLint level,
- GLint internalFormat,
- GLint width, GLint height, GLint border,
- GLenum format, GLenum type, const void *pixels,
- const struct gl_pixelstore_attrib *packing,
- struct gl_texture_object *texObj,
- struct gl_texture_image *texImage)
-{
- viaTexImage( ctx, 2, target, level,
- internalFormat, width, height, border,
- format, type, pixels,
- packing, texObj, texImage );
-}
-
-static void viaTexSubImage2D(GLcontext *ctx,
- GLenum target,
- GLint level,
- GLint xoffset, GLint yoffset,
- GLsizei width, GLsizei height,
- GLenum format, GLenum type,
- const GLvoid *pixels,
- const struct gl_pixelstore_attrib *packing,
- struct gl_texture_object *texObj,
- struct gl_texture_image *texImage)
-{
- struct via_context *vmesa = VIA_CONTEXT(ctx);
-
- viaWaitIdle(vmesa, GL_TRUE);
- vmesa->clearTexCache = 1;
-
- _mesa_store_texsubimage2d(ctx, target, level, xoffset, yoffset, width,
- height, format, type, pixels, packing, texObj,
- texImage);
-}
-
-static void viaTexImage1D(GLcontext *ctx,
- GLenum target, GLint level,
- GLint internalFormat,
- GLint width, GLint border,
- GLenum format, GLenum type, const void *pixels,
- const struct gl_pixelstore_attrib *packing,
- struct gl_texture_object *texObj,
- struct gl_texture_image *texImage)
-{
- viaTexImage( ctx, 1, target, level,
- internalFormat, width, 1, border,
- format, type, pixels,
- packing, texObj, texImage );
-}
-
-static void viaTexSubImage1D(GLcontext *ctx,
- GLenum target,
- GLint level,
- GLint xoffset,
- GLsizei width,
- GLenum format, GLenum type,
- const GLvoid *pixels,
- const struct gl_pixelstore_attrib *packing,
- struct gl_texture_object *texObj,
- struct gl_texture_image *texImage)
-{
- struct via_context *vmesa = VIA_CONTEXT(ctx);
-
- viaWaitIdle(vmesa, GL_TRUE);
- vmesa->clearTexCache = 1;
-
- _mesa_store_texsubimage1d(ctx, target, level, xoffset, width,
- format, type, pixels, packing, texObj,
- texImage);
-}
-
-
-
-static GLboolean viaIsTextureResident(GLcontext *ctx,
- struct gl_texture_object *texObj)
-{
- struct via_texture_object *viaObj =
- (struct via_texture_object *)texObj;
-
- return (viaObj->memType == VIA_MEM_AGP ||
- viaObj->memType == VIA_MEM_VIDEO);
-}
-
-
-
-static struct gl_texture_image *viaNewTextureImage( GLcontext *ctx )
-{
- (void) ctx;
- return (struct gl_texture_image *)CALLOC_STRUCT(via_texture_image);
-}
-
-
-static struct gl_texture_object *viaNewTextureObject( GLcontext *ctx,
- GLuint name,
- GLenum target )
-{
- struct via_texture_object *obj = CALLOC_STRUCT(via_texture_object);
-
- _mesa_initialize_texture_object(&obj->obj, name, target);
- (void) ctx;
-
- obj->memType = VIA_MEM_UNKNOWN;
-
- return &obj->obj;
-}
-
-
-static void viaFreeTextureImageData( GLcontext *ctx,
- struct gl_texture_image *texImage )
-{
- struct via_context *vmesa = VIA_CONTEXT(ctx);
- struct via_texture_image *image = (struct via_texture_image *)texImage;
-
- if (image->texMem) {
- via_free_texture(vmesa, image->texMem);
- image->texMem = NULL;
- }
-
- texImage->Data = NULL;
-}
-
-
-
-
-void viaInitTextureFuncs(struct dd_function_table * functions)
-{
- functions->ChooseTextureFormat = viaChooseTexFormat;
- functions->TexImage1D = viaTexImage1D;
- functions->TexImage2D = viaTexImage2D;
- functions->TexSubImage1D = viaTexSubImage1D;
- functions->TexSubImage2D = viaTexSubImage2D;
-
- functions->NewTextureObject = viaNewTextureObject;
- functions->NewTextureImage = viaNewTextureImage;
- functions->DeleteTexture = _mesa_delete_texture_object;
- functions->FreeTexImageData = viaFreeTextureImageData;
-
-#if 0 && defined( USE_SSE_ASM )
- /*
- * XXX this code is disabled for now because the via_sse_memcpy()
- * routine causes segfaults with flightgear.
- * See Mesa3d-dev mail list messages from 7/15/2005 for details.
- * Note that this function is currently disabled in via_tris.c too.
- */
- if (getenv("VIA_NO_SSE"))
- functions->TextureMemCpy = _mesa_memcpy;
- else
- functions->TextureMemCpy = via_sse_memcpy;
-#else
- functions->TextureMemCpy = _mesa_memcpy;
-#endif
-
- functions->UpdateTexturePalette = 0;
- functions->IsTextureResident = viaIsTextureResident;
-}
-
-