void
intelReadPixels(GLcontext * ctx,
GLint x, GLint y, GLsizei width, GLsizei height,
GLenum format, GLenum type,
const struct gl_pixelstore_attrib *pack, GLvoid * pixels)
{
if (INTEL_DEBUG & DEBUG_PIXEL)
fprintf(stderr, "%s\n", __FUNCTION__);
intelFlush(ctx);
intel_prepare_render(intel_context(ctx));
if (do_blit_readpixels
(ctx, x, y, width, height, format, type, pack, pixels))
return;
if (INTEL_DEBUG & DEBUG_PIXEL)
printf("%s: fallback to swrast\n", __FUNCTION__);
/* Update Mesa state before calling down into _swrast_ReadPixels, as
* the spans code requires the computed buffer states to be up to date,
* but _swrast_ReadPixels only updates Mesa state after setting up
* the spans code.
*/
if (ctx->NewState)
_mesa_update_state(ctx);
_swrast_ReadPixels(ctx, x, y, width, height, format, type, pack, pixels);
}
static bool
intel_copy_texsubimage(struct intel_context *intel,
struct intel_texture_image *intelImage,
GLint dstx, GLint dsty, GLint slice,
struct intel_renderbuffer *irb,
GLint x, GLint y, GLsizei width, GLsizei height)
{
const GLenum internalFormat = intelImage->base.Base.InternalFormat;
intel_prepare_render(intel);
if (!intelImage->mt || !irb || !irb->mt) {
if (unlikely(INTEL_DEBUG & DEBUG_PERF))
fprintf(stderr, "%s fail %p %p (0x%08x)\n",
__FUNCTION__, intelImage->mt, irb, internalFormat);
return false;
}
/* blit from src buffer to texture */
if (!intel_miptree_blit(intel,
irb->mt, irb->mt_level, irb->mt_layer,
x, y, irb->Base.Base.Name == 0,
intelImage->mt, intelImage->base.Base.Level,
intelImage->base.Base.Face + slice,
dstx, dsty, false,
width, height, GL_COPY)) {
return false;
}
return true;
}
/**
* Called by ctx->Driver.Clear.
*/
static void
brw_clear(struct gl_context *ctx, GLbitfield mask)
{
struct brw_context *brw = brw_context(ctx);
struct gl_framebuffer *fb = ctx->DrawBuffer;
bool partial_clear = ctx->Scissor.EnableFlags && !noop_scissor(ctx, fb);
if (!_mesa_check_conditional_render(ctx))
return;
if (mask & (BUFFER_BIT_FRONT_LEFT | BUFFER_BIT_FRONT_RIGHT)) {
brw->front_buffer_dirty = true;
}
intel_prepare_render(brw);
brw_workaround_depthstencil_alignment(brw, partial_clear ? 0 : mask);
if (mask & BUFFER_BIT_DEPTH) {
if (brw_fast_clear_depth(ctx)) {
DBG("fast clear: depth\n");
mask &= ~BUFFER_BIT_DEPTH;
}
}
/* BLORP is currently only supported on Gen6+. */
if (brw->gen >= 6 && brw->gen < 8) {
if (mask & BUFFER_BITS_COLOR) {
if (brw_blorp_clear_color(brw, fb, mask, partial_clear)) {
debug_mask("blorp color", mask & BUFFER_BITS_COLOR);
mask &= ~BUFFER_BITS_COLOR;
}
}
}
GLbitfield tri_mask = mask & (BUFFER_BITS_COLOR |
BUFFER_BIT_STENCIL |
BUFFER_BIT_DEPTH);
if (tri_mask) {
debug_mask("tri", tri_mask);
mask &= ~tri_mask;
if (ctx->API == API_OPENGLES) {
_mesa_meta_Clear(&brw->ctx, tri_mask);
} else {
_mesa_meta_glsl_Clear(&brw->ctx, tri_mask);
}
}
/* Any strange buffers get passed off to swrast */
if (mask) {
debug_mask("swrast", mask);
_swrast_Clear(ctx, mask);
}
}
GLboolean
intelMakeCurrent(__DRIcontext * driContextPriv,
__DRIdrawable * driDrawPriv,
__DRIdrawable * driReadPriv)
{
struct brw_context *brw;
GET_CURRENT_CONTEXT(curCtx);
if (driContextPriv)
brw = (struct brw_context *) driContextPriv->driverPrivate;
else
brw = NULL;
/* According to the glXMakeCurrent() man page: "Pending commands to
* the previous context, if any, are flushed before it is released."
* But only flush if we're actually changing contexts.
*/
if (brw_context(curCtx) && brw_context(curCtx) != brw) {
_mesa_flush(curCtx);
}
if (driContextPriv) {
struct gl_context *ctx = &brw->ctx;
struct gl_framebuffer *fb, *readFb;
if (driDrawPriv == NULL && driReadPriv == NULL) {
fb = _mesa_get_incomplete_framebuffer();
readFb = _mesa_get_incomplete_framebuffer();
} else {
fb = driDrawPriv->driverPrivate;
readFb = driReadPriv->driverPrivate;
driContextPriv->dri2.draw_stamp = driDrawPriv->dri2.stamp - 1;
driContextPriv->dri2.read_stamp = driReadPriv->dri2.stamp - 1;
}
/* The sRGB workaround changes the renderbuffer's format. We must change
* the format before the renderbuffer's miptree get's allocated, otherwise
* the formats of the renderbuffer and its miptree will differ.
*/
intel_gles3_srgb_workaround(brw, fb);
intel_gles3_srgb_workaround(brw, readFb);
/* If the context viewport hasn't been initialized, force a call out to
* the loader to get buffers so we have a drawable size for the initial
* viewport. */
if (!brw->ctx.ViewportInitialized)
intel_prepare_render(brw);
_mesa_make_current(ctx, fb, readFb);
} else {
_mesa_make_current(NULL, NULL, NULL);
}
return true;
}
static bool
intel_copy_texsubimage(struct brw_context *brw,
struct intel_texture_image *intelImage,
GLint dstx, GLint dsty, GLint slice,
struct intel_renderbuffer *irb,
GLint x, GLint y, GLsizei width, GLsizei height)
{
const GLenum internalFormat = intelImage->base.Base.InternalFormat;
bool ret;
/* No pixel transfer operations (zoom, bias, mapping), just a blit */
if (brw->ctx._ImageTransferState)
return false;
intel_prepare_render(brw);
/* glCopyTexSubImage() can be called on a multisampled renderbuffer (if
* that renderbuffer is associated with the window system framebuffer),
* however the hardware blitter can't handle this case, so fall back to
* meta (which can, since it uses ReadPixels).
*/
if (irb->Base.Base.NumSamples != 0)
return false;
/* glCopyTexSubImage() can't be called on a multisampled texture. */
assert(intelImage->base.Base.NumSamples == 0);
if (!intelImage->mt || !irb || !irb->mt) {
if (unlikely(INTEL_DEBUG & DEBUG_PERF))
fprintf(stderr, "%s fail %p %p (0x%08x)\n",
__func__, intelImage->mt, irb, internalFormat);
return false;
}
/* account for view parameters and face index */
int dst_level = intelImage->base.Base.Level +
intelImage->base.Base.TexObject->MinLevel;
int dst_slice = slice + intelImage->base.Base.Face +
intelImage->base.Base.TexObject->MinLayer;
_mesa_unlock_texture(&brw->ctx, intelImage->base.Base.TexObject);
/* blit from src buffer to texture */
ret = intel_miptree_blit(brw,
irb->mt, irb->mt_level, irb->mt_layer,
x, y, irb->Base.Base.Name == 0,
intelImage->mt, dst_level, dst_slice,
dstx, dsty, false,
width, height, GL_COPY);
_mesa_lock_texture(&brw->ctx, intelImage->base.Base.TexObject);
return ret;
}
GLboolean
intelMakeCurrent(__DRIcontext * driContextPriv,
__DRIdrawable * driDrawPriv,
__DRIdrawable * driReadPriv)
{
struct intel_context *intel;
GET_CURRENT_CONTEXT(curCtx);
if (driContextPriv)
intel = (struct intel_context *) driContextPriv->driverPrivate;
else
intel = NULL;
/* According to the glXMakeCurrent() man page: "Pending commands to
* the previous context, if any, are flushed before it is released."
* But only flush if we're actually changing contexts.
*/
if (intel_context(curCtx) && intel_context(curCtx) != intel) {
_mesa_flush(curCtx);
}
if (driContextPriv) {
struct gl_context *ctx = &intel->ctx;
struct gl_framebuffer *fb, *readFb;
if (driDrawPriv == NULL && driReadPriv == NULL) {
fb = _mesa_get_incomplete_framebuffer();
readFb = _mesa_get_incomplete_framebuffer();
} else {
fb = driDrawPriv->driverPrivate;
readFb = driReadPriv->driverPrivate;
driContextPriv->dri2.draw_stamp = driDrawPriv->dri2.stamp - 1;
driContextPriv->dri2.read_stamp = driReadPriv->dri2.stamp - 1;
}
intel_prepare_render(intel);
_mesa_make_current(ctx, fb, readFb);
intel_gles3_srgb_workaround(intel, ctx->WinSysDrawBuffer);
intel_gles3_srgb_workaround(intel, ctx->WinSysReadBuffer);
/* We do this in intel_prepare_render() too, but intel->ctx.DrawBuffer
* is NULL at that point. We can't call _mesa_makecurrent()
* first, since we need the buffer size for the initial
* viewport. So just call intel_draw_buffer() again here. */
intel_draw_buffer(ctx);
}
else {
_mesa_make_current(NULL, NULL, NULL);
}
return true;
}
/**
* Copy mipmap image between trees
*/
void
intel_miptree_image_copy(struct intel_context *intel,
struct intel_mipmap_tree *dst,
GLuint face, GLuint level,
struct intel_mipmap_tree *src)
{
GLuint width = src->level[level].width;
GLuint height = src->level[level].height;
GLuint depth = src->level[level].depth;
GLuint src_x, src_y, dst_x, dst_y;
GLuint i;
GLboolean success;
if (dst->compressed) {
GLuint align_w, align_h;
intel_get_texture_alignment_unit(dst->internal_format,
&align_w, &align_h);
height = (height + 3) / 4;
width = ALIGN(width, align_w);
}
intel_prepare_render(intel);
for (i = 0; i < depth; i++) {
intel_miptree_get_image_offset(src, level, face, i, &src_x, &src_y);
intel_miptree_get_image_offset(dst, level, face, i, &dst_x, &dst_y);
success = intel_region_copy(intel,
dst->region, 0, dst_x, dst_y,
src->region, 0, src_x, src_y,
width, height, GL_FALSE,
GL_COPY);
if (!success) {
GLubyte *src_ptr, *dst_ptr;
src_ptr = intel_region_map(intel, src->region);
dst_ptr = intel_region_map(intel, dst->region);
_mesa_copy_rect(dst_ptr,
dst->cpp,
dst->region->pitch,
dst_x, dst_y, width, height,
src_ptr,
src->region->pitch,
src_x, src_y);
intel_region_unmap(intel, src->region);
intel_region_unmap(intel, dst->region);
}
}
}
/**
* \see dd_function_table::MapRenderbuffer
*/
static void
intel_map_renderbuffer(struct gl_context *ctx,
struct gl_renderbuffer *rb,
GLuint x, GLuint y, GLuint w, GLuint h,
GLbitfield mode,
GLubyte **out_map,
GLint *out_stride)
{
struct brw_context *brw = brw_context(ctx);
struct swrast_renderbuffer *srb = (struct swrast_renderbuffer *)rb;
struct intel_renderbuffer *irb = intel_renderbuffer(rb);
void *map;
int stride;
if (srb->Buffer) {
/* this is a malloc'd renderbuffer (accum buffer), not an irb */
GLint bpp = _mesa_get_format_bytes(rb->Format);
GLint rowStride = srb->RowStride;
*out_map = (GLubyte *) srb->Buffer + y * rowStride + x * bpp;
*out_stride = rowStride;
return;
}
intel_prepare_render(brw);
/* For a window-system renderbuffer, we need to flip the mapping we receive
* upside-down. So we need to ask for a rectangle on flipped vertically, and
* we then return a pointer to the bottom of it with a negative stride.
*/
if (rb->Name == 0) {
y = rb->Height - y - h;
}
intel_miptree_map(brw, irb->mt, irb->mt_level, irb->mt_layer,
x, y, w, h, mode, &map, &stride);
if (rb->Name == 0) {
map += (h - 1) * stride;
stride = -stride;
}
DBG("%s: rb %d (%s) mt mapped: (%d, %d) (%dx%d) -> %p/%d\n",
__FUNCTION__, rb->Name, _mesa_get_format_name(rb->Format),
x, y, w, h, map, stride);
*out_map = map;
*out_stride = stride;
}
GLboolean
intelMakeCurrent(__DRIcontext * driContextPriv,
__DRIdrawable * driDrawPriv,
__DRIdrawable * driReadPriv)
{
struct brw_context *brw;
GET_CURRENT_CONTEXT(curCtx);
if (driContextPriv)
brw = (struct brw_context *) driContextPriv->driverPrivate;
else
brw = NULL;
/* According to the glXMakeCurrent() man page: "Pending commands to
* the previous context, if any, are flushed before it is released."
* But only flush if we're actually changing contexts.
*/
if (brw_context(curCtx) && brw_context(curCtx) != brw) {
_mesa_flush(curCtx);
}
if (driContextPriv) {
struct gl_context *ctx = &brw->ctx;
struct gl_framebuffer *fb, *readFb;
if (driDrawPriv == NULL && driReadPriv == NULL) {
fb = _mesa_get_incomplete_framebuffer();
readFb = _mesa_get_incomplete_framebuffer();
} else {
fb = driDrawPriv->driverPrivate;
readFb = driReadPriv->driverPrivate;
driContextPriv->dri2.draw_stamp = driDrawPriv->dri2.stamp - 1;
driContextPriv->dri2.read_stamp = driReadPriv->dri2.stamp - 1;
}
intel_prepare_render(brw);
_mesa_make_current(ctx, fb, readFb);
intel_gles3_srgb_workaround(brw, ctx->WinSysDrawBuffer);
intel_gles3_srgb_workaround(brw, ctx->WinSysReadBuffer);
}
else {
_mesa_make_current(NULL, NULL, NULL);
}
return true;
}
void
intelReadPixels(struct gl_context * ctx,
GLint x, GLint y, GLsizei width, GLsizei height,
GLenum format, GLenum type,
const struct gl_pixelstore_attrib *pack, GLvoid * pixels)
{
bool ok;
struct brw_context *brw = brw_context(ctx);
bool dirty;
DBG("%s\n", __func__);
/* Reading pixels wont dirty the front buffer, so reset the dirty
* flag after calling intel_prepare_render().
*/
dirty = brw->front_buffer_dirty;
intel_prepare_render(brw);
brw->front_buffer_dirty = dirty;
if (_mesa_is_bufferobj(pack->BufferObj)) {
if (intel_readpixels_blorp(ctx, x, y, width, height,
format, type, pixels, pack))
return;
perf_debug("%s: fallback to CPU mapping in PBO case\n", __func__);
}
ok = intel_readpixels_tiled_memcpy(ctx, x, y, width, height,
format, type, pixels, pack);
if(ok)
return;
/* Update Mesa state before calling _mesa_readpixels().
* XXX this may not be needed since ReadPixels no longer uses the
* span code.
*/
if (ctx->NewState)
_mesa_update_state(ctx);
_mesa_readpixels(ctx, x, y, width, height, format, type, pack, pixels);
/* There's an intel_prepare_render() call in intelSpanRenderStart(). */
brw->front_buffer_dirty = dirty;
}
/**
* Called by ctx->Driver.Clear.
*/
static void
brw_clear(struct gl_context *ctx, GLbitfield mask)
{
struct intel_context *intel = intel_context(ctx);
if (!_mesa_check_conditional_render(ctx))
return;
if (mask & (BUFFER_BIT_FRONT_LEFT | BUFFER_BIT_FRONT_RIGHT)) {
intel->front_buffer_dirty = true;
}
intel_prepare_render(intel);
if (mask & BUFFER_BIT_DEPTH) {
if (brw_fast_clear_depth(ctx)) {
DBG("fast clear: depth\n");
mask &= ~BUFFER_BIT_DEPTH;
}
}
GLbitfield tri_mask = mask & (BUFFER_BITS_COLOR |
BUFFER_BIT_STENCIL |
BUFFER_BIT_DEPTH);
if (tri_mask) {
debug_mask("tri", tri_mask);
mask &= ~tri_mask;
if (ctx->API == API_OPENGLES) {
_mesa_meta_Clear(&intel->ctx, tri_mask);
} else {
_mesa_meta_glsl_Clear(&intel->ctx, tri_mask);
}
}
/* Any strange buffers get passed off to swrast */
if (mask) {
debug_mask("swrast", mask);
_swrast_Clear(ctx, mask);
}
}
void
intelReadPixels(struct gl_context * ctx,
GLint x, GLint y, GLsizei width, GLsizei height,
GLenum format, GLenum type,
const struct gl_pixelstore_attrib *pack, GLvoid * pixels)
{
struct brw_context *brw = brw_context(ctx);
bool dirty;
DBG("%s\n", __FUNCTION__);
if (_mesa_is_bufferobj(pack->BufferObj)) {
/* Using PBOs, so try the BLT based path. */
if (do_blit_readpixels(ctx, x, y, width, height, format, type, pack,
pixels)) {
return;
}
perf_debug("%s: fallback to CPU mapping in PBO case\n", __FUNCTION__);
}
/* glReadPixels() wont dirty the front buffer, so reset the dirty
* flag after calling intel_prepare_render(). */
dirty = brw->front_buffer_dirty;
intel_prepare_render(brw);
brw->front_buffer_dirty = dirty;
/* Update Mesa state before calling _mesa_readpixels().
* XXX this may not be needed since ReadPixels no longer uses the
* span code.
*/
if (ctx->NewState)
_mesa_update_state(ctx);
_mesa_readpixels(ctx, x, y, width, height, format, type, pack, pixels);
/* There's an intel_prepare_render() call in intelSpanRenderStart(). */
brw->front_buffer_dirty = dirty;
}
/**
* Prepare for software rendering. Map current read/draw framebuffers'
* renderbuffes and all currently bound texture objects.
*
* Old note: Moved locking out to get reasonable span performance.
*/
void
intelSpanRenderStart(struct gl_context * ctx)
{
struct intel_context *intel = intel_context(ctx);
GLuint i;
intel_flush(&intel->ctx);
intel_prepare_render(intel);
for (i = 0; i < ctx->Const.MaxTextureImageUnits; i++) {
if (ctx->Texture.Unit[i]._ReallyEnabled) {
struct gl_texture_object *texObj = ctx->Texture.Unit[i]._Current;
intel_finalize_mipmap_tree(intel, i);
intel_tex_map_images(intel, intel_texture_object(texObj));
}
}
intel_map_unmap_framebuffer(intel, ctx->DrawBuffer, GL_TRUE);
if (ctx->ReadBuffer != ctx->DrawBuffer)
intel_map_unmap_framebuffer(intel, ctx->ReadBuffer, GL_TRUE);
}
GLboolean
intelMakeCurrent(__DRIcontext * driContextPriv,
__DRIdrawable * driDrawPriv,
__DRIdrawable * driReadPriv)
{
struct intel_context *intel;
GET_CURRENT_CONTEXT(curCtx);
if (driContextPriv)
intel = (struct intel_context *) driContextPriv->driverPrivate;
else
intel = NULL;
/* According to the glXMakeCurrent() man page: "Pending commands to
* the previous context, if any, are flushed before it is released."
* But only flush if we're actually changing contexts.
*/
if (intel_context(curCtx) && intel_context(curCtx) != intel) {
_mesa_flush(curCtx);
}
if (driContextPriv) {
struct gl_framebuffer *fb = driDrawPriv->driverPrivate;
struct gl_framebuffer *readFb = driReadPriv->driverPrivate;
intel->driReadDrawable = driReadPriv;
intel->driDrawable = driDrawPriv;
driContextPriv->dri2.draw_stamp = driDrawPriv->dri2.stamp - 1;
driContextPriv->dri2.read_stamp = driReadPriv->dri2.stamp - 1;
intel_prepare_render(intel);
_mesa_make_current(&intel->ctx, fb, readFb);
}
else {
_mesa_make_current(NULL, NULL, NULL);
}
return GL_TRUE;
}
bool
intel_copy_texsubimage(struct intel_context *intel,
struct intel_texture_image *intelImage,
GLint dstx, GLint dsty,
struct intel_renderbuffer *irb,
GLint x, GLint y, GLsizei width, GLsizei height)
{
struct gl_context *ctx = &intel->ctx;
struct intel_region *region;
const GLenum internalFormat = intelImage->base.Base.InternalFormat;
bool copy_supported = false;
bool copy_supported_with_alpha_override = false;
intel_prepare_render(intel);
if (!intelImage->mt || !irb || !irb->mt) {
if (unlikely(INTEL_DEBUG & DEBUG_PERF))
fprintf(stderr, "%s fail %p %p (0x%08x)\n",
__FUNCTION__, intelImage->mt, irb, internalFormat);
return false;
} else {
region = irb->mt->region;
assert(region);
}
/* According to the Ivy Bridge PRM, Vol1 Part4, section 1.2.1.2 (Graphics
* Data Size Limitations):
*
* The BLT engine is capable of transferring very large quantities of
* graphics data. Any graphics data read from and written to the
* destination is permitted to represent a number of pixels that
* occupies up to 65,536 scan lines and up to 32,768 bytes per scan line
* at the destination. The maximum number of pixels that may be
* represented per scan line’s worth of graphics data depends on the
* color depth.
*
* Furthermore, intelEmitCopyBlit (which is called below) uses a signed
* 16-bit integer to represent buffer pitch, so it can only handle buffer
* pitches < 32k.
*
* As a result of these two limitations, we can only use the blitter to do
* this copy when the region's pitch is less than 32k.
*/
if (region->pitch >= 32768)
return false;
if (intelImage->base.Base.TexObject->Target == GL_TEXTURE_1D_ARRAY ||
intelImage->base.Base.TexObject->Target == GL_TEXTURE_2D_ARRAY) {
perf_debug("no support for array textures\n");
}
copy_supported = intelImage->base.Base.TexFormat == intel_rb_format(irb);
/* Converting ARGB8888 to XRGB8888 is trivial: ignore the alpha bits */
if (intel_rb_format(irb) == MESA_FORMAT_ARGB8888 &&
intelImage->base.Base.TexFormat == MESA_FORMAT_XRGB8888) {
copy_supported = true;
}
/* Converting XRGB8888 to ARGB8888 requires setting the alpha bits to 1.0 */
if (intel_rb_format(irb) == MESA_FORMAT_XRGB8888 &&
intelImage->base.Base.TexFormat == MESA_FORMAT_ARGB8888) {
copy_supported_with_alpha_override = true;
}
if (!copy_supported && !copy_supported_with_alpha_override) {
if (unlikely(INTEL_DEBUG & DEBUG_PERF))
fprintf(stderr, "%s mismatched formats %s, %s\n",
__FUNCTION__,
_mesa_get_format_name(intelImage->base.Base.TexFormat),
_mesa_get_format_name(intel_rb_format(irb)));
return false;
}
{
GLuint image_x, image_y;
GLshort src_pitch;
/* get dest x/y in destination texture */
intel_miptree_get_image_offset(intelImage->mt,
intelImage->base.Base.Level,
intelImage->base.Base.Face,
&image_x, &image_y);
/* The blitter can't handle Y-tiled buffers. */
if (intelImage->mt->region->tiling == I915_TILING_Y) {
return false;
}
if (_mesa_is_winsys_fbo(ctx->ReadBuffer)) {
/* Flip vertical orientation for system framebuffers */
y = ctx->ReadBuffer->Height - (y + height);
src_pitch = -region->pitch;
} else {
/* reading from a FBO, y is already oriented the way we like */
src_pitch = region->pitch;
}
/* blit from src buffer to texture */
if (!intelEmitCopyBlit(intel,
intelImage->mt->cpp,
src_pitch,
region->bo,
0,
region->tiling,
intelImage->mt->region->pitch,
intelImage->mt->region->bo,
0,
intelImage->mt->region->tiling,
irb->draw_x + x, irb->draw_y + y,
image_x + dstx, image_y + dsty,
width, height,
GL_COPY)) {
return false;
}
}
if (copy_supported_with_alpha_override)
intel_set_teximage_alpha_to_one(ctx, intelImage);
return true;
}
/* May fail if out of video memory for texture or vbo upload, or on
* fallback conditions.
*/
static void
brw_try_draw_prims(struct gl_context *ctx,
const struct gl_client_array *arrays[],
const struct _mesa_prim *prims,
GLuint nr_prims,
const struct _mesa_index_buffer *ib,
GLuint min_index,
GLuint max_index,
struct gl_buffer_object *indirect)
{
struct brw_context *brw = brw_context(ctx);
GLuint i;
bool fail_next = false;
if (ctx->NewState)
_mesa_update_state(ctx);
/* Find the highest sampler unit used by each shader program. A bit-count
* won't work since ARB programs use the texture unit number as the sampler
* index.
*/
brw->wm.base.sampler_count =
_mesa_fls(ctx->FragmentProgram._Current->Base.SamplersUsed);
brw->gs.base.sampler_count = ctx->GeometryProgram._Current ?
_mesa_fls(ctx->GeometryProgram._Current->Base.SamplersUsed) : 0;
brw->vs.base.sampler_count =
_mesa_fls(ctx->VertexProgram._Current->Base.SamplersUsed);
/* We have to validate the textures *before* checking for fallbacks;
* otherwise, the software fallback won't be able to rely on the
* texture state, the firstLevel and lastLevel fields won't be
* set in the intel texture object (they'll both be 0), and the
* software fallback will segfault if it attempts to access any
* texture level other than level 0.
*/
brw_validate_textures(brw);
intel_prepare_render(brw);
/* This workaround has to happen outside of brw_upload_render_state()
* because it may flush the batchbuffer for a blit, affecting the state
* flags.
*/
brw_workaround_depthstencil_alignment(brw, 0);
/* Bind all inputs, derive varying and size information:
*/
brw_merge_inputs(brw, arrays);
brw->ib.ib = ib;
brw->ctx.NewDriverState |= BRW_NEW_INDICES;
brw->vb.min_index = min_index;
brw->vb.max_index = max_index;
brw->ctx.NewDriverState |= BRW_NEW_VERTICES;
for (i = 0; i < nr_prims; i++) {
int estimated_max_prim_size;
const int sampler_state_size = 16;
estimated_max_prim_size = 512; /* batchbuffer commands */
estimated_max_prim_size += BRW_MAX_TEX_UNIT *
(sampler_state_size + sizeof(struct gen5_sampler_default_color));
estimated_max_prim_size += 1024; /* gen6 VS push constants */
estimated_max_prim_size += 1024; /* gen6 WM push constants */
estimated_max_prim_size += 512; /* misc. pad */
/* Flush the batch if it's approaching full, so that we don't wrap while
* we've got validated state that needs to be in the same batch as the
* primitives.
*/
intel_batchbuffer_require_space(brw, estimated_max_prim_size, RENDER_RING);
intel_batchbuffer_save_state(brw);
if (brw->num_instances != prims[i].num_instances ||
brw->basevertex != prims[i].basevertex) {
brw->num_instances = prims[i].num_instances;
brw->basevertex = prims[i].basevertex;
if (i > 0) { /* For i == 0 we just did this before the loop */
brw->ctx.NewDriverState |= BRW_NEW_VERTICES;
brw_merge_inputs(brw, arrays);
}
}
brw->draw.gl_basevertex =
prims[i].indexed ? prims[i].basevertex : prims[i].start;
drm_intel_bo_unreference(brw->draw.draw_params_bo);
if (prims[i].is_indirect) {
/* Point draw_params_bo at the indirect buffer. */
brw->draw.draw_params_bo =
intel_buffer_object(ctx->DrawIndirectBuffer)->buffer;
drm_intel_bo_reference(brw->draw.draw_params_bo);
brw->draw.draw_params_offset =
prims[i].indirect_offset + (prims[i].indexed ? 12 : 8);
} else {
/* Set draw_params_bo to NULL so brw_prepare_vertices knows it
* has to upload gl_BaseVertex and such if they're needed.
*/
brw->draw.draw_params_bo = NULL;
brw->draw.draw_params_offset = 0;
}
if (brw->gen < 6)
brw_set_prim(brw, &prims[i]);
else
gen6_set_prim(brw, &prims[i]);
retry:
/* Note that before the loop, brw->ctx.NewDriverState was set to != 0, and
* that the state updated in the loop outside of this block is that in
* *_set_prim or intel_batchbuffer_flush(), which only impacts
* brw->ctx.NewDriverState.
*/
if (brw->ctx.NewDriverState) {
brw->no_batch_wrap = true;
brw_upload_render_state(brw);
}
brw_emit_prim(brw, &prims[i], brw->primitive);
brw->no_batch_wrap = false;
if (dri_bufmgr_check_aperture_space(&brw->batch.bo, 1)) {
if (!fail_next) {
intel_batchbuffer_reset_to_saved(brw);
intel_batchbuffer_flush(brw);
fail_next = true;
goto retry;
} else {
int ret = intel_batchbuffer_flush(brw);
WARN_ONCE(ret == -ENOSPC,
"i965: Single primitive emit exceeded "
"available aperture space\n");
}
}
/* Now that we know we haven't run out of aperture space, we can safely
* reset the dirty bits.
*/
if (brw->ctx.NewDriverState)
brw_render_state_finished(brw);
}
if (brw->always_flush_batch)
intel_batchbuffer_flush(brw);
brw_state_cache_check_size(brw);
brw_postdraw_set_buffers_need_resolve(brw);
return;
}
/* May fail if out of video memory for texture or vbo upload, or on
* fallback conditions.
*/
static bool brw_try_draw_prims( struct gl_context *ctx,
const struct gl_client_array *arrays[],
const struct _mesa_prim *prim,
GLuint nr_prims,
const struct _mesa_index_buffer *ib,
GLuint min_index,
GLuint max_index )
{
struct intel_context *intel = intel_context(ctx);
struct brw_context *brw = brw_context(ctx);
bool retval = true;
GLuint i;
bool fail_next = false;
if (ctx->NewState)
_mesa_update_state( ctx );
/* We have to validate the textures *before* checking for fallbacks;
* otherwise, the software fallback won't be able to rely on the
* texture state, the firstLevel and lastLevel fields won't be
* set in the intel texture object (they'll both be 0), and the
* software fallback will segfault if it attempts to access any
* texture level other than level 0.
*/
brw_validate_textures( brw );
/* Resolves must occur after updating state and finalizing textures but
* before setting up any hardware state for this draw call.
*/
brw_predraw_resolve_buffers(brw);
/* Bind all inputs, derive varying and size information:
*/
brw_merge_inputs( brw, arrays );
brw->ib.ib = ib;
brw->state.dirty.brw |= BRW_NEW_INDICES;
brw->vb.min_index = min_index;
brw->vb.max_index = max_index;
brw->state.dirty.brw |= BRW_NEW_VERTICES;
/* Have to validate state quite late. Will rebuild tnl_program,
* which depends on varying information.
*
* Note this is where brw->vs->prog_data.inputs_read is calculated,
* so can't access it earlier.
*/
intel_prepare_render(intel);
for (i = 0; i < nr_prims; i++) {
int estimated_max_prim_size;
estimated_max_prim_size = 512; /* batchbuffer commands */
estimated_max_prim_size += (BRW_MAX_TEX_UNIT *
(sizeof(struct brw_sampler_state) +
sizeof(struct gen5_sampler_default_color)));
estimated_max_prim_size += 1024; /* gen6 VS push constants */
estimated_max_prim_size += 1024; /* gen6 WM push constants */
estimated_max_prim_size += 512; /* misc. pad */
/* Flush the batch if it's approaching full, so that we don't wrap while
* we've got validated state that needs to be in the same batch as the
* primitives.
*/
intel_batchbuffer_require_space(intel, estimated_max_prim_size, false);
intel_batchbuffer_save_state(intel);
if (intel->gen < 6)
brw_set_prim(brw, &prim[i]);
else
gen6_set_prim(brw, &prim[i]);
retry:
/* Note that before the loop, brw->state.dirty.brw was set to != 0, and
* that the state updated in the loop outside of this block is that in
* *_set_prim or intel_batchbuffer_flush(), which only impacts
* brw->state.dirty.brw.
*/
if (brw->state.dirty.brw) {
intel->no_batch_wrap = true;
brw_upload_state(brw);
if (unlikely(brw->intel.Fallback)) {
intel->no_batch_wrap = false;
retval = false;
goto out;
}
}
if (intel->gen >= 7)
gen7_emit_prim(brw, &prim[i], brw->primitive);
else
brw_emit_prim(brw, &prim[i], brw->primitive);
intel->no_batch_wrap = false;
if (dri_bufmgr_check_aperture_space(&intel->batch.bo, 1)) {
if (!fail_next) {
intel_batchbuffer_reset_to_saved(intel);
intel_batchbuffer_flush(intel);
fail_next = true;
goto retry;
} else {
if (intel_batchbuffer_flush(intel) == -ENOSPC) {
static bool warned = false;
if (!warned) {
fprintf(stderr, "i965: Single primitive emit exceeded"
"available aperture space\n");
warned = true;
}
retval = false;
}
}
}
if (!_mesa_meta_in_progress(ctx))
brw_update_primitive_count(brw, &prim[i]);
}
if (intel->always_flush_batch)
intel_batchbuffer_flush(intel);
out:
brw_state_cache_check_size(brw);
brw_postdraw_set_buffers_need_resolve(brw);
return retval;
}
static void
i915_render_start(struct intel_context *intel)
{
intel_prepare_render(intel);
}
void
intelReadPixels(struct gl_context * ctx,
GLint x, GLint y, GLsizei width, GLsizei height,
GLenum format, GLenum type,
const struct gl_pixelstore_attrib *pack, GLvoid * pixels)
{
bool ok;
struct brw_context *brw = brw_context(ctx);
bool dirty;
DBG("%s\n", __func__);
if (_mesa_is_bufferobj(pack->BufferObj)) {
if (_mesa_meta_pbo_GetTexSubImage(ctx, 2, NULL, x, y, 0, width, height, 1,
format, type, pixels, pack)) {
/* _mesa_meta_pbo_GetTexSubImage() implements PBO transfers by
* binding the user-provided BO as a fake framebuffer and rendering
* to it. This breaks the invariant of the GL that nothing is able
* to render to a BO, causing nondeterministic corruption issues
* because the render cache is not coherent with a number of other
* caches that the BO could potentially be bound to afterwards.
*
* This could be solved in the same way that we guarantee texture
* coherency after a texture is attached to a framebuffer and
* rendered to, but that would involve checking *all* BOs bound to
* the pipeline for the case we need to emit a cache flush due to
* previous rendering to any of them -- Including vertex, index,
* uniform, atomic counter, shader image, transform feedback,
* indirect draw buffers, etc.
*
* That would increase the per-draw call overhead even though it's
* very unlikely that any of the BOs bound to the pipeline has been
* rendered to via a PBO at any point, so it seems better to just
* flush here unconditionally.
*/
brw_emit_mi_flush(brw);
return;
}
perf_debug("%s: fallback to CPU mapping in PBO case\n", __func__);
}
ok = intel_readpixels_tiled_memcpy(ctx, x, y, width, height,
format, type, pixels, pack);
if(ok)
return;
/* glReadPixels() wont dirty the front buffer, so reset the dirty
* flag after calling intel_prepare_render(). */
dirty = brw->front_buffer_dirty;
intel_prepare_render(brw);
brw->front_buffer_dirty = dirty;
/* Update Mesa state before calling _mesa_readpixels().
* XXX this may not be needed since ReadPixels no longer uses the
* span code.
*/
if (ctx->NewState)
_mesa_update_state(ctx);
_mesa_readpixels(ctx, x, y, width, height, format, type, pack, pixels);
/* There's an intel_prepare_render() call in intelSpanRenderStart(). */
brw->front_buffer_dirty = dirty;
}
/**
* Called by ctx->Driver.Clear.
*/
static void
intelClear(struct gl_context *ctx, GLbitfield mask)
{
struct intel_context *intel = intel_context(ctx);
const GLuint colorMask = *((GLuint *) & ctx->Color.ColorMask[0]);
GLbitfield tri_mask = 0;
GLbitfield blit_mask = 0;
GLbitfield swrast_mask = 0;
struct gl_framebuffer *fb = ctx->DrawBuffer;
struct intel_renderbuffer *irb;
int i;
if (!_mesa_check_conditional_render(ctx))
return;
if (mask & (BUFFER_BIT_FRONT_LEFT | BUFFER_BIT_FRONT_RIGHT)) {
intel->front_buffer_dirty = GL_TRUE;
}
if (0)
fprintf(stderr, "%s\n", __FUNCTION__);
/* Get SW clears out of the way: Anything without an intel_renderbuffer */
for (i = 0; i < BUFFER_COUNT; i++) {
if (!(mask & (1 << i)))
continue;
irb = intel_get_renderbuffer(fb, i);
if (unlikely(!irb)) {
swrast_mask |= (1 << i);
mask &= ~(1 << i);
}
}
if (unlikely(swrast_mask)) {
debug_mask("swrast", swrast_mask);
_swrast_Clear(ctx, swrast_mask);
}
/* HW color buffers (front, back, aux, generic FBO, etc) */
if (intel->gen < 6 && colorMask == ~0) {
/* clear all R,G,B,A */
blit_mask |= (mask & BUFFER_BITS_COLOR);
}
else {
/* glColorMask in effect */
tri_mask |= (mask & BUFFER_BITS_COLOR);
}
/* Make sure we have up to date buffers before we start looking at
* the tiling bits to determine how to clear. */
intel_prepare_render(intel);
/* HW stencil */
if (mask & BUFFER_BIT_STENCIL) {
const struct intel_region *stencilRegion
= intel_get_rb_region(fb, BUFFER_STENCIL);
if (stencilRegion) {
/* have hw stencil */
if (stencilRegion->tiling == I915_TILING_Y ||
(ctx->Stencil.WriteMask[0] & 0xff) != 0xff) {
/* We have to use the 3D engine if we're clearing a partial mask
* of the stencil buffer, or if we're on a 965 which has a tiled
* depth/stencil buffer in a layout we can't blit to.
*/
tri_mask |= BUFFER_BIT_STENCIL;
}
else if (intel->has_separate_stencil &&
stencilRegion->tiling == I915_TILING_NONE) {
/* The stencil buffer is actually W tiled, which the hardware
* cannot blit to. */
tri_mask |= BUFFER_BIT_STENCIL;
}
else {
/* clearing all stencil bits, use blitting */
blit_mask |= BUFFER_BIT_STENCIL;
}
}
}
/* HW depth */
if (mask & BUFFER_BIT_DEPTH) {
const struct intel_region *irb = intel_get_rb_region(fb, BUFFER_DEPTH);
/* clear depth with whatever method is used for stencil (see above) */
if (irb->tiling == I915_TILING_Y || tri_mask & BUFFER_BIT_STENCIL)
tri_mask |= BUFFER_BIT_DEPTH;
else
blit_mask |= BUFFER_BIT_DEPTH;
}
/* If we're doing a tri pass for depth/stencil, include a likely color
* buffer with it.
*/
if (mask & (BUFFER_BIT_DEPTH | BUFFER_BIT_STENCIL)) {
int color_bit = _mesa_ffs(mask & BUFFER_BITS_COLOR);
if (color_bit != 0) {
tri_mask |= blit_mask & (1 << (color_bit - 1));
blit_mask &= ~(1 << (color_bit - 1));
}
}
/* Anything left, just use tris */
tri_mask |= mask & ~blit_mask;
if (blit_mask) {
debug_mask("blit", blit_mask);
tri_mask |= intelClearWithBlit(ctx, blit_mask);
}
if (tri_mask) {
debug_mask("tri", tri_mask);
if (ctx->Extensions.ARB_fragment_shader)
_mesa_meta_glsl_Clear(&intel->ctx, tri_mask);
else
_mesa_meta_Clear(&intel->ctx, tri_mask);
}
}
bool
intel_copy_texsubimage(struct intel_context *intel,
struct intel_texture_image *intelImage,
GLint dstx, GLint dsty,
struct intel_renderbuffer *irb,
GLint x, GLint y, GLsizei width, GLsizei height)
{
struct gl_context *ctx = &intel->ctx;
struct intel_region *region;
const GLenum internalFormat = intelImage->base.Base.InternalFormat;
bool copy_supported = false;
bool copy_supported_with_alpha_override = false;
intel_prepare_render(intel);
if (!intelImage->mt || !irb || !irb->mt) {
if (unlikely(INTEL_DEBUG & DEBUG_FALLBACKS))
fprintf(stderr, "%s fail %p %p (0x%08x)\n",
__FUNCTION__, intelImage->mt, irb, internalFormat);
return false;
} else {
region = irb->mt->region;
assert(region);
}
copy_supported = intelImage->base.Base.TexFormat == intel_rb_format(irb);
/* Converting ARGB8888 to XRGB8888 is trivial: ignore the alpha bits */
if (intel_rb_format(irb) == MESA_FORMAT_ARGB8888 &&
intelImage->base.Base.TexFormat == MESA_FORMAT_XRGB8888) {
copy_supported = true;
}
/* Converting XRGB8888 to ARGB8888 requires setting the alpha bits to 1.0 */
if (intel_rb_format(irb) == MESA_FORMAT_XRGB8888 &&
intelImage->base.Base.TexFormat == MESA_FORMAT_ARGB8888) {
copy_supported_with_alpha_override = true;
}
if (!copy_supported && !copy_supported_with_alpha_override) {
if (unlikely(INTEL_DEBUG & DEBUG_FALLBACKS))
fprintf(stderr, "%s mismatched formats %s, %s\n",
__FUNCTION__,
_mesa_get_format_name(intelImage->base.Base.TexFormat),
_mesa_get_format_name(intel_rb_format(irb)));
return false;
}
{
GLuint image_x, image_y;
GLshort src_pitch;
/* get dest x/y in destination texture */
intel_miptree_get_image_offset(intelImage->mt,
intelImage->base.Base.Level,
intelImage->base.Base.Face,
0,
&image_x, &image_y);
/* The blitter can't handle Y-tiled buffers. */
if (intelImage->mt->region->tiling == I915_TILING_Y) {
return false;
}
if (ctx->ReadBuffer->Name == 0) {
/* Flip vertical orientation for system framebuffers */
y = ctx->ReadBuffer->Height - (y + height);
src_pitch = -region->pitch;
} else {
/* reading from a FBO, y is already oriented the way we like */
src_pitch = region->pitch;
}
/* blit from src buffer to texture */
if (!intelEmitCopyBlit(intel,
intelImage->mt->cpp,
src_pitch,
region->bo,
0,
region->tiling,
intelImage->mt->region->pitch,
intelImage->mt->region->bo,
0,
intelImage->mt->region->tiling,
irb->draw_x + x, irb->draw_y + y,
image_x + dstx, image_y + dsty,
width, height,
GL_COPY)) {
return false;
}
}
if (copy_supported_with_alpha_override)
intel_set_teximage_alpha_to_one(ctx, intelImage);
return true;
}
/**
* CopyPixels with the blitter. Don't support zooming, pixel transfer, etc.
*/
static GLboolean
do_blit_copypixels(struct gl_context * ctx,
GLint srcx, GLint srcy,
GLsizei width, GLsizei height,
GLint dstx, GLint dsty, GLenum type)
{
struct intel_context *intel = intel_context(ctx);
struct gl_framebuffer *fb = ctx->DrawBuffer;
struct gl_framebuffer *read_fb = ctx->ReadBuffer;
GLint orig_dstx;
GLint orig_dsty;
GLint orig_srcx;
GLint orig_srcy;
GLboolean flip = GL_FALSE;
struct intel_renderbuffer *draw_irb = NULL;
struct intel_renderbuffer *read_irb = NULL;
/* Update draw buffer bounds */
_mesa_update_state(ctx);
switch (type) {
case GL_COLOR:
if (fb->_NumColorDrawBuffers != 1) {
fallback_debug("glCopyPixels() fallback: MRT\n");
return GL_FALSE;
}
draw_irb = intel_renderbuffer(fb->_ColorDrawBuffers[0]);
read_irb = intel_renderbuffer(read_fb->_ColorReadBuffer);
break;
case GL_DEPTH_STENCIL_EXT:
draw_irb = intel_renderbuffer(fb->Attachment[BUFFER_DEPTH].Renderbuffer);
read_irb =
intel_renderbuffer(read_fb->Attachment[BUFFER_DEPTH].Renderbuffer);
break;
case GL_DEPTH:
fallback_debug("glCopyPixels() fallback: GL_DEPTH\n");
return GL_FALSE;
case GL_STENCIL:
fallback_debug("glCopyPixels() fallback: GL_STENCIL\n");
return GL_FALSE;
default:
fallback_debug("glCopyPixels(): Unknown type\n");
return GL_FALSE;
}
if (!draw_irb) {
fallback_debug("glCopyPixels() fallback: missing draw buffer\n");
return GL_FALSE;
}
if (!read_irb) {
fallback_debug("glCopyPixels() fallback: missing read buffer\n");
return GL_FALSE;
}
if (draw_irb->Base.Format != read_irb->Base.Format &&
!(draw_irb->Base.Format == MESA_FORMAT_XRGB8888 &&
read_irb->Base.Format == MESA_FORMAT_ARGB8888)) {
fallback_debug("glCopyPixels() fallback: mismatched formats (%s -> %s\n",
_mesa_get_format_name(read_irb->Base.Format),
_mesa_get_format_name(draw_irb->Base.Format));
return GL_FALSE;
}
/* Copypixels can be more than a straight copy. Ensure all the
* extra operations are disabled:
*/
if (!intel_check_copypixel_blit_fragment_ops(ctx) ||
ctx->Pixel.ZoomX != 1.0F || ctx->Pixel.ZoomY != 1.0F)
return GL_FALSE;
intel_prepare_render(intel);
intel_flush(&intel->ctx);
/* Clip to destination buffer. */
orig_dstx = dstx;
orig_dsty = dsty;
if (!_mesa_clip_to_region(fb->_Xmin, fb->_Ymin,
fb->_Xmax, fb->_Ymax,
&dstx, &dsty, &width, &height))
goto out;
/* Adjust src coords for our post-clipped destination origin */
srcx += dstx - orig_dstx;
srcy += dsty - orig_dsty;
/* Clip to source buffer. */
orig_srcx = srcx;
orig_srcy = srcy;
if (!_mesa_clip_to_region(0, 0,
read_fb->Width, read_fb->Height,
&srcx, &srcy, &width, &height))
goto out;
/* Adjust dst coords for our post-clipped source origin */
dstx += srcx - orig_srcx;
dsty += srcy - orig_srcy;
/* Flip dest Y if it's a window system framebuffer. */
if (fb->Name == 0) {
/* copypixels to a window system framebuffer */
dsty = fb->Height - dsty - height;
flip = !flip;
}
/* Flip source Y if it's a window system framebuffer. */
if (read_fb->Name == 0) {
srcy = read_fb->Height - srcy - height;
flip = !flip;
}
srcx += read_irb->draw_x;
srcy += read_irb->draw_y;
dstx += draw_irb->draw_x;
dsty += draw_irb->draw_y;
if (!intel_region_copy(intel,
draw_irb->region, 0, dstx, dsty,
read_irb->region, 0, srcx, srcy,
width, height, flip,
ctx->Color.ColorLogicOpEnabled ?
ctx->Color.LogicOp : GL_COPY)) {
DBG("%s: blit failure\n", __FUNCTION__);
return GL_FALSE;
}
out:
intel_check_front_buffer_rendering(intel);
DBG("%s: success\n", __FUNCTION__);
return GL_TRUE;
}
/*
* Render a bitmap.
*/
static bool
do_blit_bitmap( struct gl_context *ctx,
GLint dstx, GLint dsty,
GLsizei width, GLsizei height,
const struct gl_pixelstore_attrib *unpack,
const GLubyte *bitmap )
{
struct intel_context *intel = intel_context(ctx);
struct gl_framebuffer *fb = ctx->DrawBuffer;
struct intel_renderbuffer *irb;
GLfloat tmpColor[4];
GLubyte ubcolor[4];
GLuint color;
GLsizei bitmap_width = width;
GLsizei bitmap_height = height;
GLint px, py;
GLuint stipple[32];
GLint orig_dstx = dstx;
GLint orig_dsty = dsty;
/* Update draw buffer bounds */
_mesa_update_state(ctx);
if (ctx->Depth.Test) {
/* The blit path produces incorrect results when depth testing is on.
* It seems the blit Z coord is always 1.0 (the far plane) so fragments
* will likely be obscured by other, closer geometry.
*/
return false;
}
intel_prepare_render(intel);
if (fb->_NumColorDrawBuffers != 1) {
perf_debug("accelerated glBitmap() only supports rendering to a "
"single color buffer\n");
return false;
}
irb = intel_renderbuffer(fb->_ColorDrawBuffers[0]);
if (_mesa_is_bufferobj(unpack->BufferObj)) {
bitmap = map_pbo(ctx, width, height, unpack, bitmap);
if (bitmap == NULL)
return true; /* even though this is an error, we're done */
}
COPY_4V(tmpColor, ctx->Current.RasterColor);
if (_mesa_need_secondary_color(ctx)) {
ADD_3V(tmpColor, tmpColor, ctx->Current.RasterSecondaryColor);
}
UNCLAMPED_FLOAT_TO_UBYTE(ubcolor[0], tmpColor[0]);
UNCLAMPED_FLOAT_TO_UBYTE(ubcolor[1], tmpColor[1]);
UNCLAMPED_FLOAT_TO_UBYTE(ubcolor[2], tmpColor[2]);
UNCLAMPED_FLOAT_TO_UBYTE(ubcolor[3], tmpColor[3]);
switch (irb->mt->format) {
case MESA_FORMAT_B8G8R8A8_UNORM:
case MESA_FORMAT_B8G8R8X8_UNORM:
color = PACK_COLOR_8888(ubcolor[3], ubcolor[0], ubcolor[1], ubcolor[2]);
break;
case MESA_FORMAT_B5G6R5_UNORM:
color = PACK_COLOR_565(ubcolor[0], ubcolor[1], ubcolor[2]);
break;
default:
perf_debug("Unsupported format %s in accelerated glBitmap()\n",
_mesa_get_format_name(irb->mt->format));
return false;
}
if (!intel_check_blit_fragment_ops(ctx, tmpColor[3] == 1.0F))
return false;
/* Clip to buffer bounds and scissor. */
if (!_mesa_clip_to_region(fb->_Xmin, fb->_Ymin,
fb->_Xmax, fb->_Ymax,
&dstx, &dsty, &width, &height))
goto out;
dsty = y_flip(fb, dsty, height);
#define DY 32
#define DX 32
/* Chop it all into chunks that can be digested by hardware: */
for (py = 0; py < height; py += DY) {
for (px = 0; px < width; px += DX) {
int h = MIN2(DY, height - py);
int w = MIN2(DX, width - px);
GLuint sz = ALIGN(ALIGN(w,8) * h, 64)/8;
GLenum logic_op = ctx->Color.ColorLogicOpEnabled ?
ctx->Color.LogicOp : GL_COPY;
assert(sz <= sizeof(stipple));
memset(stipple, 0, sz);
/* May need to adjust this when padding has been introduced in
* sz above:
*
* Have to translate destination coordinates back into source
* coordinates.
*/
int count = get_bitmap_rect(bitmap_width, bitmap_height, unpack,
bitmap,
-orig_dstx + (dstx + px),
-orig_dsty + y_flip(fb, dsty + py, h),
w, h,
(GLubyte *)stipple,
8,
_mesa_is_winsys_fbo(fb));
if (count == 0)
continue;
if (!intelEmitImmediateColorExpandBlit(intel,
irb->mt->cpp,
(GLubyte *)stipple,
sz,
color,
irb->mt->region->pitch,
irb->mt->region->bo,
0,
irb->mt->region->tiling,
dstx + px,
dsty + py,
w, h,
logic_op)) {
return false;
}
if (ctx->Query.CurrentOcclusionObject)
ctx->Query.CurrentOcclusionObject->Result += count;
}
}
out:
if (unlikely(INTEL_DEBUG & DEBUG_SYNC))
intel_batchbuffer_flush(intel);
if (_mesa_is_bufferobj(unpack->BufferObj)) {
/* done with PBO so unmap it now */
ctx->Driver.UnmapBuffer(ctx, unpack->BufferObj, MAP_INTERNAL);
}
intel_check_front_buffer_rendering(intel);
return true;
}
/* May fail if out of video memory for texture or vbo upload, or on
* fallback conditions.
*/
static GLboolean brw_try_draw_prims( struct gl_context *ctx,
const struct gl_client_array *arrays[],
const struct _mesa_prim *prim,
GLuint nr_prims,
const struct _mesa_index_buffer *ib,
GLuint min_index,
GLuint max_index )
{
struct intel_context *intel = intel_context(ctx);
struct brw_context *brw = brw_context(ctx);
GLboolean retval = GL_FALSE;
GLboolean warn = GL_FALSE;
GLuint i;
if (ctx->NewState)
_mesa_update_state( ctx );
/* We have to validate the textures *before* checking for fallbacks;
* otherwise, the software fallback won't be able to rely on the
* texture state, the firstLevel and lastLevel fields won't be
* set in the intel texture object (they'll both be 0), and the
* software fallback will segfault if it attempts to access any
* texture level other than level 0.
*/
brw_validate_textures( brw );
/* Bind all inputs, derive varying and size information:
*/
brw_merge_inputs( brw, arrays );
brw->ib.ib = ib;
brw->state.dirty.brw |= BRW_NEW_INDICES;
brw->vb.min_index = min_index;
brw->vb.max_index = max_index;
brw->state.dirty.brw |= BRW_NEW_VERTICES;
/* Have to validate state quite late. Will rebuild tnl_program,
* which depends on varying information.
*
* Note this is where brw->vs->prog_data.inputs_read is calculated,
* so can't access it earlier.
*/
intel_prepare_render(intel);
for (i = 0; i < nr_prims; i++) {
uint32_t hw_prim;
int estimated_max_prim_size;
estimated_max_prim_size = 512; /* batchbuffer commands */
estimated_max_prim_size += (BRW_MAX_TEX_UNIT *
(sizeof(struct brw_sampler_state) +
sizeof(struct gen5_sampler_default_color)));
estimated_max_prim_size += 1024; /* gen6 VS push constants */
estimated_max_prim_size += 1024; /* gen6 WM push constants */
estimated_max_prim_size += 512; /* misc. pad */
/* Flush the batch if it's approaching full, so that we don't wrap while
* we've got validated state that needs to be in the same batch as the
* primitives.
*/
intel_batchbuffer_require_space(intel, estimated_max_prim_size, false);
hw_prim = brw_set_prim(brw, &prim[i]);
if (brw->state.dirty.brw) {
brw_validate_state(brw);
/* Various fallback checks: */
if (brw->intel.Fallback)
goto out;
/* Check that we can fit our state in with our existing batchbuffer, or
* flush otherwise.
*/
if (dri_bufmgr_check_aperture_space(brw->state.validated_bos,
brw->state.validated_bo_count)) {
static GLboolean warned;
intel_batchbuffer_flush(intel);
/* Validate the state after we flushed the batch (which would have
* changed the set of dirty state). If we still fail to
* check_aperture, warn of what's happening, but attempt to continue
* on since it may succeed anyway, and the user would probably rather
* see a failure and a warning than a fallback.
*/
brw_validate_state(brw);
if (!warned &&
dri_bufmgr_check_aperture_space(brw->state.validated_bos,
brw->state.validated_bo_count)) {
warn = GL_TRUE;
warned = GL_TRUE;
}
}
intel->no_batch_wrap = GL_TRUE;
brw_upload_state(brw);
}
if (intel->gen >= 7)
gen7_emit_prim(brw, &prim[i], hw_prim);
else
brw_emit_prim(brw, &prim[i], hw_prim);
intel->no_batch_wrap = GL_FALSE;
retval = GL_TRUE;
}
if (intel->always_flush_batch)
intel_batchbuffer_flush(intel);
out:
brw_state_cache_check_size(brw);
if (warn)
fprintf(stderr, "i965: Single primitive emit potentially exceeded "
"available aperture space\n");
if (!retval)
DBG("%s failed\n", __FUNCTION__);
return retval;
}
/* May fail if out of video memory for texture or vbo upload, or on
* fallback conditions.
*/
static bool brw_try_draw_prims( struct gl_context *ctx,
const struct gl_client_array *arrays[],
const struct _mesa_prim *prims,
GLuint nr_prims,
const struct _mesa_index_buffer *ib,
GLuint min_index,
GLuint max_index,
struct gl_buffer_object *indirect)
{
struct brw_context *brw = brw_context(ctx);
bool retval = true;
GLuint i;
bool fail_next = false;
if (ctx->NewState)
_mesa_update_state( ctx );
/* Find the highest sampler unit used by each shader program. A bit-count
* won't work since ARB programs use the texture unit number as the sampler
* index.
*/
brw->wm.base.sampler_count =
_mesa_fls(ctx->FragmentProgram._Current->Base.SamplersUsed);
brw->gs.base.sampler_count = ctx->GeometryProgram._Current ?
_mesa_fls(ctx->GeometryProgram._Current->Base.SamplersUsed) : 0;
brw->vs.base.sampler_count =
_mesa_fls(ctx->VertexProgram._Current->Base.SamplersUsed);
/* We have to validate the textures *before* checking for fallbacks;
* otherwise, the software fallback won't be able to rely on the
* texture state, the firstLevel and lastLevel fields won't be
* set in the intel texture object (they'll both be 0), and the
* software fallback will segfault if it attempts to access any
* texture level other than level 0.
*/
brw_validate_textures( brw );
intel_prepare_render(brw);
/* This workaround has to happen outside of brw_upload_state() because it
* may flush the batchbuffer for a blit, affecting the state flags.
*/
brw_workaround_depthstencil_alignment(brw, 0);
/* Resolves must occur after updating renderbuffers, updating context state,
* and finalizing textures but before setting up any hardware state for
* this draw call.
*/
brw_predraw_resolve_buffers(brw);
/* Bind all inputs, derive varying and size information:
*/
brw_merge_inputs( brw, arrays );
brw->ib.ib = ib;
brw->state.dirty.brw |= BRW_NEW_INDICES;
brw->vb.min_index = min_index;
brw->vb.max_index = max_index;
brw->state.dirty.brw |= BRW_NEW_VERTICES;
for (i = 0; i < nr_prims; i++) {
int estimated_max_prim_size;
estimated_max_prim_size = 512; /* batchbuffer commands */
estimated_max_prim_size += (BRW_MAX_TEX_UNIT *
(sizeof(struct brw_sampler_state) +
sizeof(struct gen5_sampler_default_color)));
estimated_max_prim_size += 1024; /* gen6 VS push constants */
estimated_max_prim_size += 1024; /* gen6 WM push constants */
estimated_max_prim_size += 512; /* misc. pad */
/* Flush the batch if it's approaching full, so that we don't wrap while
* we've got validated state that needs to be in the same batch as the
* primitives.
*/
intel_batchbuffer_require_space(brw, estimated_max_prim_size, RENDER_RING);
intel_batchbuffer_save_state(brw);
if (brw->num_instances != prims[i].num_instances) {
brw->num_instances = prims[i].num_instances;
brw->state.dirty.brw |= BRW_NEW_VERTICES;
brw_merge_inputs(brw, arrays);
}
if (brw->basevertex != prims[i].basevertex) {
brw->basevertex = prims[i].basevertex;
brw->state.dirty.brw |= BRW_NEW_VERTICES;
brw_merge_inputs(brw, arrays);
}
if (brw->gen < 6)
brw_set_prim(brw, &prims[i]);
else
gen6_set_prim(brw, &prims[i]);
retry:
/* Note that before the loop, brw->state.dirty.brw was set to != 0, and
* that the state updated in the loop outside of this block is that in
* *_set_prim or intel_batchbuffer_flush(), which only impacts
* brw->state.dirty.brw.
*/
if (brw->state.dirty.brw) {
brw->no_batch_wrap = true;
brw_upload_state(brw);
}
brw_emit_prim(brw, &prims[i], brw->primitive);
brw->no_batch_wrap = false;
if (dri_bufmgr_check_aperture_space(&brw->batch.bo, 1)) {
if (!fail_next) {
intel_batchbuffer_reset_to_saved(brw);
intel_batchbuffer_flush(brw);
fail_next = true;
goto retry;
} else {
if (intel_batchbuffer_flush(brw) == -ENOSPC) {
static bool warned = false;
if (!warned) {
fprintf(stderr, "i965: Single primitive emit exceeded"
"available aperture space\n");
warned = true;
}
retval = false;
}
}
}
}
if (brw->always_flush_batch)
intel_batchbuffer_flush(brw);
brw_state_cache_check_size(brw);
brw_postdraw_set_buffers_need_resolve(brw);
return retval;
}
static GLboolean
do_blit_readpixels(GLcontext * ctx,
GLint x, GLint y, GLsizei width, GLsizei height,
GLenum format, GLenum type,
const struct gl_pixelstore_attrib *pack, GLvoid * pixels)
{
struct intel_context *intel = intel_context(ctx);
struct intel_region *src = intel_readbuf_region(intel);
struct intel_buffer_object *dst = intel_buffer_object(pack->BufferObj);
GLuint dst_offset;
GLuint rowLength;
drm_intel_bo *dst_buffer;
GLboolean all;
GLint dst_x, dst_y;
if (INTEL_DEBUG & DEBUG_PIXEL)
printf("%s\n", __FUNCTION__);
if (!src)
return GL_FALSE;
if (!_mesa_is_bufferobj(pack->BufferObj)) {
/* PBO only for now:
*/
if (INTEL_DEBUG & DEBUG_PIXEL)
printf("%s - not PBO\n", __FUNCTION__);
return GL_FALSE;
}
if (ctx->_ImageTransferState ||
!intel_check_blit_format(src, format, type)) {
if (INTEL_DEBUG & DEBUG_PIXEL)
printf("%s - bad format for blit\n", __FUNCTION__);
return GL_FALSE;
}
if (pack->Alignment != 1 || pack->SwapBytes || pack->LsbFirst) {
if (INTEL_DEBUG & DEBUG_PIXEL)
printf("%s: bad packing params\n", __FUNCTION__);
return GL_FALSE;
}
if (pack->RowLength > 0)
rowLength = pack->RowLength;
else
rowLength = width;
if (pack->Invert) {
if (INTEL_DEBUG & DEBUG_PIXEL)
printf("%s: MESA_PACK_INVERT not done yet\n", __FUNCTION__);
return GL_FALSE;
}
else {
if (ctx->ReadBuffer->Name == 0)
rowLength = -rowLength;
}
dst_offset = (GLintptr) _mesa_image_address(2, pack, pixels, width, height,
format, type, 0, 0, 0);
if (!_mesa_clip_copytexsubimage(ctx,
&dst_x, &dst_y,
&x, &y,
&width, &height)) {
return GL_TRUE;
}
intel_prepare_render(intel);
all = (width * height * src->cpp == dst->Base.Size &&
x == 0 && dst_offset == 0);
dst_x = 0;
dst_y = 0;
dst_buffer = intel_bufferobj_buffer(intel, dst,
all ? INTEL_WRITE_FULL :
INTEL_WRITE_PART);
if (ctx->ReadBuffer->Name == 0)
y = ctx->ReadBuffer->Height - (y + height);
if (!intelEmitCopyBlit(intel,
src->cpp,
src->pitch, src->buffer, 0, src->tiling,
rowLength, dst_buffer, dst_offset, GL_FALSE,
x, y,
dst_x, dst_y,
width, height,
GL_COPY)) {
return GL_FALSE;
}
if (INTEL_DEBUG & DEBUG_PIXEL)
printf("%s - DONE\n", __FUNCTION__);
return GL_TRUE;
}
/**
* CopyPixels with the blitter. Don't support zooming, pixel transfer, etc.
*/
static bool
do_blit_copypixels(struct gl_context * ctx,
GLint srcx, GLint srcy,
GLsizei width, GLsizei height,
GLint dstx, GLint dsty, GLenum type)
{
struct brw_context *brw = brw_context(ctx);
struct gl_framebuffer *fb = ctx->DrawBuffer;
struct gl_framebuffer *read_fb = ctx->ReadBuffer;
GLint orig_dstx;
GLint orig_dsty;
GLint orig_srcx;
GLint orig_srcy;
struct intel_renderbuffer *draw_irb = NULL;
struct intel_renderbuffer *read_irb = NULL;
/* Update draw buffer bounds */
_mesa_update_state(ctx);
intel_prepare_render(brw);
switch (type) {
case GL_COLOR:
if (fb->_NumColorDrawBuffers != 1) {
perf_debug("glCopyPixels() fallback: MRT\n");
return false;
}
draw_irb = intel_renderbuffer(fb->_ColorDrawBuffers[0]);
read_irb = intel_renderbuffer(read_fb->_ColorReadBuffer);
break;
case GL_DEPTH_STENCIL_EXT:
draw_irb = intel_renderbuffer(fb->Attachment[BUFFER_DEPTH].Renderbuffer);
read_irb =
intel_renderbuffer(read_fb->Attachment[BUFFER_DEPTH].Renderbuffer);
break;
case GL_DEPTH:
perf_debug("glCopyPixels() fallback: GL_DEPTH\n");
return false;
case GL_STENCIL:
perf_debug("glCopyPixels() fallback: GL_STENCIL\n");
return false;
default:
perf_debug("glCopyPixels(): Unknown type\n");
return false;
}
if (!draw_irb) {
perf_debug("glCopyPixels() fallback: missing draw buffer\n");
return false;
}
if (!read_irb) {
perf_debug("glCopyPixels() fallback: missing read buffer\n");
return false;
}
if (draw_irb->mt->num_samples > 1 || read_irb->mt->num_samples > 1) {
perf_debug("glCopyPixels() fallback: multisampled buffers\n");
return false;
}
if (ctx->_ImageTransferState) {
perf_debug("glCopyPixels(): Unsupported image transfer state\n");
return false;
}
if (ctx->Depth.Test) {
perf_debug("glCopyPixels(): Unsupported depth test state\n");
return false;
}
if (ctx->Stencil._Enabled) {
perf_debug("glCopyPixels(): Unsupported stencil test state\n");
return false;
}
if (ctx->Fog.Enabled ||
ctx->Texture._MaxEnabledTexImageUnit != -1 ||
ctx->FragmentProgram._Enabled) {
perf_debug("glCopyPixels(): Unsupported fragment shader state\n");
return false;
}
if (ctx->Color.AlphaEnabled ||
ctx->Color.BlendEnabled) {
perf_debug("glCopyPixels(): Unsupported blend state\n");
return false;
}
if (!ctx->Color.ColorMask[0][0] ||
!ctx->Color.ColorMask[0][1] ||
!ctx->Color.ColorMask[0][2] ||
!ctx->Color.ColorMask[0][3]) {
perf_debug("glCopyPixels(): Unsupported color mask state\n");
return false;
}
if (ctx->Pixel.ZoomX != 1.0F || ctx->Pixel.ZoomY != 1.0F) {
perf_debug("glCopyPixles(): Unsupported pixel zoom\n");
return false;
}
intel_batchbuffer_flush(brw);
/* Clip to destination buffer. */
orig_dstx = dstx;
orig_dsty = dsty;
if (!_mesa_clip_to_region(fb->_Xmin, fb->_Ymin,
fb->_Xmax, fb->_Ymax,
&dstx, &dsty, &width, &height))
goto out;
/* Adjust src coords for our post-clipped destination origin */
srcx += dstx - orig_dstx;
srcy += dsty - orig_dsty;
/* Clip to source buffer. */
orig_srcx = srcx;
orig_srcy = srcy;
if (!_mesa_clip_to_region(0, 0,
read_fb->Width, read_fb->Height,
&srcx, &srcy, &width, &height))
goto out;
/* Adjust dst coords for our post-clipped source origin */
dstx += srcx - orig_srcx;
dsty += srcy - orig_srcy;
if (!intel_miptree_blit(brw,
read_irb->mt, read_irb->mt_level, read_irb->mt_layer,
srcx, srcy, _mesa_is_winsys_fbo(read_fb),
draw_irb->mt, draw_irb->mt_level, draw_irb->mt_layer,
dstx, dsty, _mesa_is_winsys_fbo(fb),
width, height,
(ctx->Color.ColorLogicOpEnabled ?
ctx->Color.LogicOp : GL_COPY))) {
DBG("%s: blit failure\n", __FUNCTION__);
return false;
}
if (ctx->Query.CurrentOcclusionObject)
ctx->Query.CurrentOcclusionObject->Result += width * height;
out:
DBG("%s: success\n", __FUNCTION__);
return true;
}
/**
* Try to do a glBlitFramebuffer using glCopyTexSubImage2D
* We can do this when the dst renderbuffer is actually a texture and
* there is no scaling, mirroring or scissoring.
*
* \return new buffer mask indicating the buffers left to blit using the
* normal path.
*/
static GLbitfield
intel_blit_framebuffer_with_blitter(struct gl_context *ctx,
GLint srcX0, GLint srcY0,
GLint srcX1, GLint srcY1,
GLint dstX0, GLint dstY0,
GLint dstX1, GLint dstY1,
GLbitfield mask, GLenum filter)
{
struct brw_context *brw = brw_context(ctx);
/* Sync up the state of window system buffers. We need to do this before
* we go looking for the buffers.
*/
intel_prepare_render(brw);
if (mask & GL_COLOR_BUFFER_BIT) {
GLint i;
const struct gl_framebuffer *drawFb = ctx->DrawBuffer;
const struct gl_framebuffer *readFb = ctx->ReadBuffer;
struct gl_renderbuffer *src_rb = readFb->_ColorReadBuffer;
struct intel_renderbuffer *src_irb = intel_renderbuffer(src_rb);
if (!src_irb) {
perf_debug("glBlitFramebuffer(): missing src renderbuffer. "
"Falling back to software rendering.\n");
return mask;
}
/* If the source and destination are the same size with no mirroring,
* the rectangles are within the size of the texture and there is no
* scissor, then we can probably use the blit engine.
*/
if (!(srcX0 - srcX1 == dstX0 - dstX1 &&
srcY0 - srcY1 == dstY0 - dstY1 &&
srcX1 >= srcX0 &&
srcY1 >= srcY0 &&
srcX0 >= 0 && srcX1 <= readFb->Width &&
srcY0 >= 0 && srcY1 <= readFb->Height &&
dstX0 >= 0 && dstX1 <= drawFb->Width &&
dstY0 >= 0 && dstY1 <= drawFb->Height &&
!ctx->Scissor.Enabled)) {
perf_debug("glBlitFramebuffer(): non-1:1 blit. "
"Falling back to software rendering.\n");
return mask;
}
/* Blit to all active draw buffers. We don't do any pre-checking,
* because we assume that copying to MRTs is rare, and failure midway
* through copying is even more rare. Even if it was to occur, it's
* safe to let meta start the copy over from scratch, because
* glBlitFramebuffer completely overwrites the destination pixels, and
* results are undefined if any destination pixels have a dependency on
* source pixels.
*/
for (i = 0; i < ctx->DrawBuffer->_NumColorDrawBuffers; i++) {
struct gl_renderbuffer *dst_rb = ctx->DrawBuffer->_ColorDrawBuffers[i];
struct intel_renderbuffer *dst_irb = intel_renderbuffer(dst_rb);
if (!dst_irb) {
perf_debug("glBlitFramebuffer(): missing dst renderbuffer. "
"Falling back to software rendering.\n");
return mask;
}
gl_format src_format = _mesa_get_srgb_format_linear(src_rb->Format);
gl_format dst_format = _mesa_get_srgb_format_linear(dst_rb->Format);
if (src_format != dst_format) {
perf_debug("glBlitFramebuffer(): unsupported blit from %s to %s. "
"Falling back to software rendering.\n",
_mesa_get_format_name(src_format),
_mesa_get_format_name(dst_format));
return mask;
}
if (!intel_miptree_blit(brw,
src_irb->mt,
src_irb->mt_level, src_irb->mt_layer,
srcX0, srcY0, src_rb->Name == 0,
dst_irb->mt,
dst_irb->mt_level, dst_irb->mt_layer,
dstX0, dstY0, dst_rb->Name == 0,
dstX1 - dstX0, dstY1 - dstY0, GL_COPY)) {
perf_debug("glBlitFramebuffer(): unknown blit failure. "
"Falling back to software rendering.\n");
return mask;
}
}
mask &= ~GL_COLOR_BUFFER_BIT;
}
return mask;
}
bool
brw_blorp_copytexsubimage(struct brw_context *brw,
struct gl_renderbuffer *src_rb,
struct gl_texture_image *dst_image,
int slice,
int srcX0, int srcY0,
int dstX0, int dstY0,
int width, int height)
{
struct gl_context *ctx = &brw->ctx;
struct intel_renderbuffer *src_irb = intel_renderbuffer(src_rb);
struct intel_texture_image *intel_image = intel_texture_image(dst_image);
/* No pixel transfer operations (zoom, bias, mapping), just a blit */
if (brw->ctx._ImageTransferState)
return false;
/* Sync up the state of window system buffers. We need to do this before
* we go looking at the src renderbuffer's miptree.
*/
intel_prepare_render(brw);
struct intel_mipmap_tree *src_mt = src_irb->mt;
struct intel_mipmap_tree *dst_mt = intel_image->mt;
/* There is support for only up to eight samples. */
if (src_mt->num_samples > 8 || dst_mt->num_samples > 8)
return false;
/* BLORP is only supported from Gen6 onwards. */
if (brw->gen < 6)
return false;
if (_mesa_get_format_base_format(src_rb->Format) !=
_mesa_get_format_base_format(dst_image->TexFormat)) {
return false;
}
/* We can't handle format conversions between Z24 and other formats since
* we have to lie about the surface format. See the comments in
* brw_blorp_surface_info::set().
*/
if ((src_mt->format == MESA_FORMAT_Z24_UNORM_X8_UINT) !=
(dst_mt->format == MESA_FORMAT_Z24_UNORM_X8_UINT)) {
return false;
}
if (!brw->format_supported_as_render_target[dst_image->TexFormat])
return false;
/* Source clipping shouldn't be necessary, since copytexsubimage (in
* src/mesa/main/teximage.c) calls _mesa_clip_copytexsubimage() which
* takes care of it.
*
* Destination clipping shouldn't be necessary since the restrictions on
* glCopyTexSubImage prevent the user from specifying a destination rectangle
* that falls outside the bounds of the destination texture.
* See error_check_subtexture_dimensions().
*/
int srcY1 = srcY0 + height;
int srcX1 = srcX0 + width;
int dstX1 = dstX0 + width;
int dstY1 = dstY0 + height;
/* Account for the fact that in the system framebuffer, the origin is at
* the lower left.
*/
bool mirror_y = false;
if (_mesa_is_winsys_fbo(ctx->ReadBuffer)) {
GLint tmp = src_rb->Height - srcY0;
srcY0 = src_rb->Height - srcY1;
srcY1 = tmp;
mirror_y = true;
}
/* Account for face selection and texture view MinLayer */
int dst_slice = slice + dst_image->TexObject->MinLayer + dst_image->Face;
int dst_level = dst_image->Level + dst_image->TexObject->MinLevel;
brw_blorp_blit_miptrees(brw,
src_mt, src_irb->mt_level, src_irb->mt_layer,
src_rb->Format, blorp_get_texture_swizzle(src_irb),
dst_mt, dst_level, dst_slice,
dst_image->TexFormat,
srcX0, srcY0, srcX1, srcY1,
dstX0, dstY0, dstX1, dstY1,
GL_NEAREST, false, mirror_y,
false, false);
/* If we're copying to a packed depth stencil texture and the source
* framebuffer has separate stencil, we need to also copy the stencil data
* over.
*/
src_rb = ctx->ReadBuffer->Attachment[BUFFER_STENCIL].Renderbuffer;
if (_mesa_get_format_bits(dst_image->TexFormat, GL_STENCIL_BITS) > 0 &&
src_rb != NULL) {
src_irb = intel_renderbuffer(src_rb);
src_mt = src_irb->mt;
if (src_mt->stencil_mt)
src_mt = src_mt->stencil_mt;
if (dst_mt->stencil_mt)
dst_mt = dst_mt->stencil_mt;
if (src_mt != dst_mt) {
brw_blorp_blit_miptrees(brw,
src_mt, src_irb->mt_level, src_irb->mt_layer,
src_mt->format,
blorp_get_texture_swizzle(src_irb),
dst_mt, dst_level, dst_slice,
dst_mt->format,
srcX0, srcY0, srcX1, srcY1,
dstX0, dstY0, dstX1, dstY1,
GL_NEAREST, false, mirror_y,
false, false);
}
}
return true;
}
static bool
do_blit_drawpixels(struct gl_context * ctx,
GLint x, GLint y, GLsizei width, GLsizei height,
GLenum format, GLenum type,
const struct gl_pixelstore_attrib *unpack,
const GLvoid * pixels)
{
struct brw_context *brw = brw_context(ctx);
struct intel_buffer_object *src = intel_buffer_object(unpack->BufferObj);
GLuint src_offset;
drm_intel_bo *src_buffer;
DBG("%s\n", __FUNCTION__);
if (!intel_check_blit_fragment_ops(ctx, false))
return false;
if (ctx->DrawBuffer->_NumColorDrawBuffers != 1) {
DBG("%s: fallback due to MRT\n", __FUNCTION__);
return false;
}
struct gl_renderbuffer *rb = ctx->DrawBuffer->_ColorDrawBuffers[0];
struct intel_renderbuffer *irb = intel_renderbuffer(rb);
if (!_mesa_format_matches_format_and_type(irb->mt->format, format, type,
false)) {
DBG("%s: bad format for blit\n", __FUNCTION__);
return false;
}
if (unpack->SwapBytes || unpack->LsbFirst ||
unpack->SkipPixels || unpack->SkipRows) {
DBG("%s: bad packing params\n", __FUNCTION__);
return false;
}
int src_stride = _mesa_image_row_stride(unpack, width, format, type);
bool src_flip = false;
/* Mesa flips the src_stride for unpack->Invert, but we want our mt to have
* a normal src_stride.
*/
if (unpack->Invert) {
src_stride = -src_stride;
src_flip = true;
}
src_offset = (GLintptr)pixels;
src_offset += _mesa_image_offset(2, unpack, width, height,
format, type, 0, 0, 0);
intel_prepare_render(brw);
src_buffer = intel_bufferobj_buffer(brw, src,
src_offset, width * height *
irb->mt->cpp);
struct intel_mipmap_tree *pbo_mt =
intel_miptree_create_for_bo(brw,
src_buffer,
irb->mt->format,
src_offset,
width, height,
src_stride, I915_TILING_NONE);
if (!pbo_mt)
return false;
if (!intel_miptree_blit(brw,
pbo_mt, 0, 0,
0, 0, src_flip,
irb->mt, irb->mt_level, irb->mt_layer,
x, y, _mesa_is_winsys_fbo(ctx->DrawBuffer),
width, height, GL_COPY)) {
DBG("%s: blit failed\n", __FUNCTION__);
intel_miptree_release(&pbo_mt);
return false;
}
intel_miptree_release(&pbo_mt);
if (ctx->Query.CurrentOcclusionObject)
ctx->Query.CurrentOcclusionObject->Result += width * height;
intel_check_front_buffer_rendering(brw);
DBG("%s: success\n", __FUNCTION__);
return true;
}