static void
intel_update_image_buffers(struct intel_context *intel, __DRIdrawable *drawable)
{
struct gl_framebuffer *fb = drawable->driverPrivate;
__DRIscreen *screen = intel->intelScreen->driScrnPriv;
struct intel_renderbuffer *front_rb;
struct intel_renderbuffer *back_rb;
struct __DRIimageList images;
unsigned int format;
uint32_t buffer_mask = 0;
int ret;
front_rb = intel_get_renderbuffer(fb, BUFFER_FRONT_LEFT);
back_rb = intel_get_renderbuffer(fb, BUFFER_BACK_LEFT);
if (back_rb)
format = intel_rb_format(back_rb);
else if (front_rb)
format = intel_rb_format(front_rb);
else
return;
if (front_rb && (_mesa_is_front_buffer_drawing(fb) ||
_mesa_is_front_buffer_reading(fb) || !back_rb)) {
buffer_mask |= __DRI_IMAGE_BUFFER_FRONT;
}
if (back_rb)
buffer_mask |= __DRI_IMAGE_BUFFER_BACK;
ret = screen->image.loader->getBuffers(drawable,
driGLFormatToImageFormat(format),
&drawable->dri2.stamp,
drawable->loaderPrivate,
buffer_mask,
&images);
if (!ret)
return;
if (images.image_mask & __DRI_IMAGE_BUFFER_FRONT) {
drawable->w = images.front->width;
drawable->h = images.front->height;
intel_update_image_buffer(intel,
drawable,
front_rb,
images.front,
__DRI_IMAGE_BUFFER_FRONT);
}
if (images.image_mask & __DRI_IMAGE_BUFFER_BACK) {
drawable->w = images.back->width;
drawable->h = images.back->height;
intel_update_image_buffer(intel,
drawable,
back_rb,
images.back,
__DRI_IMAGE_BUFFER_BACK);
}
}
uint32_t
brw_depthbuffer_format(struct brw_context *brw)
{
struct intel_context *intel = &brw->intel;
struct gl_context *ctx = &intel->ctx;
struct gl_framebuffer *fb = ctx->DrawBuffer;
struct intel_renderbuffer *drb = intel_get_renderbuffer(fb, BUFFER_DEPTH);
struct intel_renderbuffer *srb;
if (!drb &&
(srb = intel_get_renderbuffer(fb, BUFFER_STENCIL)) &&
!srb->mt->stencil_mt &&
(intel_rb_format(srb) == MESA_FORMAT_S8_Z24 ||
intel_rb_format(srb) == MESA_FORMAT_Z32_FLOAT_X24S8)) {
drb = srb;
}
if (!drb)
return BRW_DEPTHFORMAT_D32_FLOAT;
switch (drb->mt->format) {
case MESA_FORMAT_Z16:
return BRW_DEPTHFORMAT_D16_UNORM;
case MESA_FORMAT_Z32_FLOAT:
return BRW_DEPTHFORMAT_D32_FLOAT;
case MESA_FORMAT_X8_Z24:
if (intel->gen >= 6) {
return BRW_DEPTHFORMAT_D24_UNORM_X8_UINT;
} else {
/* Use D24_UNORM_S8, not D24_UNORM_X8.
*
* D24_UNORM_X8 was not introduced until Gen5. (See the Ironlake PRM,
* Volume 2, Part 1, Section 8.4.6 "Depth/Stencil Buffer State", Bits
* 3DSTATE_DEPTH_BUFFER.Surface_Format).
*
* However, on Gen5, D24_UNORM_X8 may be used only if separate
* stencil is enabled, and we never enable it. From the Ironlake PRM,
* same section as above, Bit 3DSTATE_DEPTH_BUFFER.Separate_Stencil_Buffer_Enable:
* If this field is disabled, the Surface Format of the depth
* buffer cannot be D24_UNORM_X8_UINT.
*/
return BRW_DEPTHFORMAT_D24_UNORM_S8_UINT;
}
case MESA_FORMAT_S8_Z24:
return BRW_DEPTHFORMAT_D24_UNORM_S8_UINT;
case MESA_FORMAT_Z32_FLOAT_X24S8:
return BRW_DEPTHFORMAT_D32_FLOAT_S8X24_UINT;
default:
_mesa_problem(ctx, "Unexpected depth format %s\n",
_mesa_get_format_name(intel_rb_format(drb)));
return BRW_DEPTHFORMAT_D16_UNORM;
}
}
static void
intel_update_image_buffer(struct intel_context *intel,
__DRIdrawable *drawable,
struct intel_renderbuffer *rb,
__DRIimage *buffer,
enum __DRIimageBufferMask buffer_type)
{
struct intel_region *region = buffer->region;
if (!rb || !region)
return;
unsigned num_samples = rb->Base.Base.NumSamples;
if (rb->mt &&
rb->mt->region &&
rb->mt->region == region)
return;
intel_miptree_release(&rb->mt);
rb->mt = intel_miptree_create_for_image_buffer(intel,
buffer_type,
intel_rb_format(rb),
num_samples,
region);
}
/**
* \brief Assign a DRI buffer's DRM region to a renderbuffer.
*
* This is called from intel_update_renderbuffers().
*
* \par Note:
* DRI buffers whose attachment point is DRI2BufferStencil or
* DRI2BufferDepthStencil are handled as special cases.
*
* \param buffer_name is a human readable name, such as "dri2 front buffer",
* that is passed to intel_region_alloc_for_handle().
*
* \see intel_update_renderbuffers()
* \see intel_region_alloc_for_handle()
*/
static void
intel_process_dri2_buffer(struct intel_context *intel,
__DRIdrawable *drawable,
__DRIbuffer *buffer,
struct intel_renderbuffer *rb,
const char *buffer_name)
{
struct intel_region *region = NULL;
if (!rb)
return;
unsigned num_samples = rb->Base.Base.NumSamples;
/* We try to avoid closing and reopening the same BO name, because the first
* use of a mapping of the buffer involves a bunch of page faulting which is
* moderately expensive.
*/
if (num_samples == 0) {
if (rb->mt &&
rb->mt->region &&
rb->mt->region->name == buffer->name)
return;
} else {
if (rb->mt &&
rb->mt->singlesample_mt &&
rb->mt->singlesample_mt->region &&
rb->mt->singlesample_mt->region->name == buffer->name)
return;
}
if (unlikely(INTEL_DEBUG & DEBUG_DRI)) {
fprintf(stderr,
"attaching buffer %d, at %d, cpp %d, pitch %d\n",
buffer->name, buffer->attachment,
buffer->cpp, buffer->pitch);
}
intel_miptree_release(&rb->mt);
region = intel_region_alloc_for_handle(intel->intelScreen,
buffer->cpp,
drawable->w,
drawable->h,
buffer->pitch / buffer->cpp,
buffer->name,
buffer_name);
if (!region)
return;
rb->mt = intel_miptree_create_for_dri2_buffer(intel,
buffer->attachment,
intel_rb_format(rb),
num_samples,
region);
intel_region_release(®ion);
}
uint32_t
brw_depthbuffer_format(struct brw_context *brw)
{
struct gl_context *ctx = &brw->ctx;
struct gl_framebuffer *fb = ctx->DrawBuffer;
struct intel_renderbuffer *drb = intel_get_renderbuffer(fb, BUFFER_DEPTH);
struct intel_renderbuffer *srb;
if (!drb &&
(srb = intel_get_renderbuffer(fb, BUFFER_STENCIL)) &&
!srb->mt->stencil_mt &&
(intel_rb_format(srb) == MESA_FORMAT_Z24_UNORM_S8_UINT ||
intel_rb_format(srb) == MESA_FORMAT_Z32_FLOAT_S8X24_UINT)) {
drb = srb;
}
if (!drb)
return BRW_DEPTHFORMAT_D32_FLOAT;
return brw_depth_format(brw, drb->mt->format);
}
/**
* Use blitting to clear the renderbuffers named by 'flags'.
* Note: we can't use the ctx->DrawBuffer->_ColorDrawBufferIndexes field
* since that might include software renderbuffers or renderbuffers
* which we're clearing with triangles.
* \param mask bitmask of BUFFER_BIT_* values indicating buffers to clear
*/
GLbitfield
intelClearWithBlit(struct gl_context *ctx, GLbitfield mask)
{
struct intel_context *intel = intel_context(ctx);
struct gl_framebuffer *fb = ctx->DrawBuffer;
GLuint clear_depth_value, clear_depth_mask;
GLint cx, cy, cw, ch;
GLbitfield fail_mask = 0;
BATCH_LOCALS;
/* Note: we don't use this function on Gen7+ hardware, so we can safely
* ignore fast color clear issues.
*/
assert(intel->gen < 7);
/*
* Compute values for clearing the buffers.
*/
clear_depth_value = 0;
clear_depth_mask = 0;
if (mask & BUFFER_BIT_DEPTH) {
clear_depth_value = (GLuint) (fb->_DepthMax * ctx->Depth.Clear);
clear_depth_mask = XY_BLT_WRITE_RGB;
}
if (mask & BUFFER_BIT_STENCIL) {
clear_depth_value |= (ctx->Stencil.Clear & 0xff) << 24;
clear_depth_mask |= XY_BLT_WRITE_ALPHA;
}
cx = fb->_Xmin;
if (_mesa_is_winsys_fbo(fb))
cy = ctx->DrawBuffer->Height - fb->_Ymax;
else
cy = fb->_Ymin;
cw = fb->_Xmax - fb->_Xmin;
ch = fb->_Ymax - fb->_Ymin;
if (cw == 0 || ch == 0)
return 0;
/* Loop over all renderbuffers */
mask &= (1 << BUFFER_COUNT) - 1;
while (mask) {
GLuint buf = ffs(mask) - 1;
bool is_depth_stencil = buf == BUFFER_DEPTH || buf == BUFFER_STENCIL;
struct intel_renderbuffer *irb;
int x1, y1, x2, y2;
uint32_t clear_val;
uint32_t BR13, CMD;
struct intel_region *region;
int pitch, cpp;
drm_intel_bo *aper_array[2];
mask &= ~(1 << buf);
irb = intel_get_renderbuffer(fb, buf);
if (irb && irb->mt) {
region = irb->mt->region;
assert(region);
assert(region->bo);
} else {
fail_mask |= 1 << buf;
continue;
}
/* OK, clear this renderbuffer */
x1 = cx + irb->draw_x;
y1 = cy + irb->draw_y;
x2 = cx + cw + irb->draw_x;
y2 = cy + ch + irb->draw_y;
pitch = region->pitch;
cpp = region->cpp;
DBG("%s dst:buf(%p)/%d %d,%d sz:%dx%d\n",
__FUNCTION__,
region->bo, pitch,
x1, y1, x2 - x1, y2 - y1);
BR13 = 0xf0 << 16;
CMD = XY_COLOR_BLT_CMD;
/* Setup the blit command */
if (cpp == 4) {
if (is_depth_stencil) {
CMD |= clear_depth_mask;
} else {
/* clearing RGBA */
CMD |= XY_BLT_WRITE_ALPHA | XY_BLT_WRITE_RGB;
}
}
assert(region->tiling != I915_TILING_Y);
BR13 |= pitch;
if (is_depth_stencil) {
clear_val = clear_depth_value;
} else {
uint8_t clear[4];
GLfloat *color = ctx->Color.ClearColor.f;
_mesa_unclamped_float_rgba_to_ubyte(clear, color);
switch (intel_rb_format(irb)) {
case MESA_FORMAT_ARGB8888:
case MESA_FORMAT_XRGB8888:
clear_val = PACK_COLOR_8888(clear[3], clear[0],
clear[1], clear[2]);
break;
case MESA_FORMAT_RGB565:
clear_val = PACK_COLOR_565(clear[0], clear[1], clear[2]);
break;
case MESA_FORMAT_ARGB4444:
clear_val = PACK_COLOR_4444(clear[3], clear[0],
clear[1], clear[2]);
break;
case MESA_FORMAT_ARGB1555:
clear_val = PACK_COLOR_1555(clear[3], clear[0],
clear[1], clear[2]);
break;
case MESA_FORMAT_A8:
clear_val = PACK_COLOR_8888(clear[3], clear[3],
clear[3], clear[3]);
break;
default:
fail_mask |= 1 << buf;
continue;
}
}
BR13 |= br13_for_cpp(cpp);
assert(x1 < x2);
assert(y1 < y2);
/* do space check before going any further */
aper_array[0] = intel->batch.bo;
aper_array[1] = region->bo;
if (drm_intel_bufmgr_check_aperture_space(aper_array,
ARRAY_SIZE(aper_array)) != 0) {
intel_batchbuffer_flush(intel);
}
BEGIN_BATCH(6);
OUT_BATCH(CMD | (6 - 2));
OUT_BATCH(BR13);
OUT_BATCH((y1 << 16) | x1);
OUT_BATCH((y2 << 16) | x2);
OUT_RELOC_FENCED(region->bo,
I915_GEM_DOMAIN_RENDER, I915_GEM_DOMAIN_RENDER,
0);
OUT_BATCH(clear_val);
ADVANCE_BATCH();
if (intel->always_flush_cache)
intel_batchbuffer_emit_mi_flush(intel);
if (buf == BUFFER_DEPTH || buf == BUFFER_STENCIL)
mask &= ~(BUFFER_BIT_DEPTH | BUFFER_BIT_STENCIL);
}
return fail_mask;
}
static unsigned
intel_bits_per_pixel(const struct intel_renderbuffer *rb)
{
return _mesa_get_format_bytes(intel_rb_format(rb)) * 8;
}
/**
* Do additional "completeness" testing of a framebuffer object.
*/
static void
intel_validate_framebuffer(struct gl_context *ctx, struct gl_framebuffer *fb)
{
struct intel_context *intel = intel_context(ctx);
const struct intel_renderbuffer *depthRb =
intel_get_renderbuffer(fb, BUFFER_DEPTH);
const struct intel_renderbuffer *stencilRb =
intel_get_renderbuffer(fb, BUFFER_STENCIL);
struct intel_mipmap_tree *depth_mt = NULL, *stencil_mt = NULL;
int i;
DBG("%s() on fb %p (%s)\n", __FUNCTION__,
fb, (fb == ctx->DrawBuffer ? "drawbuffer" :
(fb == ctx->ReadBuffer ? "readbuffer" : "other buffer")));
if (depthRb)
depth_mt = depthRb->mt;
if (stencilRb) {
stencil_mt = stencilRb->mt;
if (stencil_mt->stencil_mt)
stencil_mt = stencil_mt->stencil_mt;
}
if (depth_mt && stencil_mt) {
if (depth_mt == stencil_mt) {
/* For true packed depth/stencil (not faked on prefers-separate-stencil
* hardware) we need to be sure they're the same level/layer, since
* we'll be emitting a single packet describing the packed setup.
*/
if (depthRb->mt_level != stencilRb->mt_level ||
depthRb->mt_layer != stencilRb->mt_layer) {
DBG("depth image level/layer %d/%d != stencil image %d/%d\n",
depthRb->mt_level,
depthRb->mt_layer,
stencilRb->mt_level,
stencilRb->mt_layer);
fb->_Status = GL_FRAMEBUFFER_UNSUPPORTED_EXT;
}
} else {
if (!intel->has_separate_stencil) {
DBG("separate stencil unsupported\n");
fb->_Status = GL_FRAMEBUFFER_UNSUPPORTED_EXT;
}
if (stencil_mt->format != MESA_FORMAT_S8) {
DBG("separate stencil is %s instead of S8\n",
_mesa_get_format_name(stencil_mt->format));
fb->_Status = GL_FRAMEBUFFER_UNSUPPORTED_EXT;
}
if (intel->gen < 7 && depth_mt->hiz_mt == NULL) {
/* Before Gen7, separate depth and stencil buffers can be used
* only if HiZ is enabled. From the Sandybridge PRM, Volume 2,
* Part 1, Bit 3DSTATE_DEPTH_BUFFER.SeparateStencilBufferEnable:
* [DevSNB]: This field must be set to the same value (enabled
* or disabled) as Hierarchical Depth Buffer Enable.
*/
DBG("separate stencil without HiZ\n");
fb->_Status = GL_FRAMEBUFFER_UNSUPPORTED;
}
}
}
for (i = 0; i < Elements(fb->Attachment); i++) {
struct gl_renderbuffer *rb;
struct intel_renderbuffer *irb;
if (fb->Attachment[i].Type == GL_NONE)
continue;
/* A supported attachment will have a Renderbuffer set either
* from being a Renderbuffer or being a texture that got the
* intel_wrap_texture() treatment.
*/
rb = fb->Attachment[i].Renderbuffer;
if (rb == NULL) {
DBG("attachment without renderbuffer\n");
fb->_Status = GL_FRAMEBUFFER_UNSUPPORTED_EXT;
continue;
}
if (fb->Attachment[i].Type == GL_TEXTURE) {
const struct gl_texture_image *img =
_mesa_get_attachment_teximage_const(&fb->Attachment[i]);
if (img->Border) {
DBG("texture with border\n");
fb->_Status = GL_FRAMEBUFFER_UNSUPPORTED_EXT;
continue;
}
}
irb = intel_renderbuffer(rb);
if (irb == NULL) {
DBG("software rendering renderbuffer\n");
fb->_Status = GL_FRAMEBUFFER_UNSUPPORTED_EXT;
continue;
}
if (!intel->vtbl.render_target_supported(intel, rb)) {
DBG("Unsupported HW texture/renderbuffer format attached: %s\n",
_mesa_get_format_name(intel_rb_format(irb)));
fb->_Status = GL_FRAMEBUFFER_UNSUPPORTED_EXT;
}
}
}
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;
}
/**
* Sets up a surface state structure to point at the given region.
* While it is only used for the front/back buffer currently, it should be
* usable for further buffers when doing ARB_draw_buffer support.
*/
static void
gen6_update_renderbuffer_surface(struct brw_context *brw,
struct gl_renderbuffer *rb,
bool layered,
unsigned int unit)
{
struct gl_context *ctx = &brw->ctx;
struct intel_renderbuffer *irb = intel_renderbuffer(rb);
struct intel_mipmap_tree *mt = irb->mt;
uint32_t *surf;
uint32_t format = 0;
/* _NEW_BUFFERS */
mesa_format rb_format = _mesa_get_render_format(ctx, intel_rb_format(irb));
uint32_t surftype;
int depth = MAX2(irb->layer_count, 1);
const GLenum gl_target =
rb->TexImage ? rb->TexImage->TexObject->Target : GL_TEXTURE_2D;
uint32_t surf_index =
brw->wm.prog_data->binding_table.render_target_start + unit;
intel_miptree_used_for_rendering(irb->mt);
surf = brw_state_batch(brw, AUB_TRACE_SURFACE_STATE, 6 * 4, 32,
&brw->wm.base.surf_offset[surf_index]);
format = brw->render_target_format[rb_format];
if (unlikely(!brw->format_supported_as_render_target[rb_format])) {
_mesa_problem(ctx, "%s: renderbuffer format %s unsupported\n",
__func__, _mesa_get_format_name(rb_format));
}
switch (gl_target) {
case GL_TEXTURE_CUBE_MAP_ARRAY:
case GL_TEXTURE_CUBE_MAP:
surftype = BRW_SURFACE_2D;
depth *= 6;
break;
case GL_TEXTURE_3D:
depth = MAX2(irb->mt->logical_depth0, 1);
/* fallthrough */
default:
surftype = translate_tex_target(gl_target);
break;
}
const int min_array_element = layered ? 0 : irb->mt_layer;
surf[0] = SET_FIELD(surftype, BRW_SURFACE_TYPE) |
SET_FIELD(format, BRW_SURFACE_FORMAT);
/* reloc */
assert(mt->offset % mt->cpp == 0);
surf[1] = mt->bo->offset64 + mt->offset;
/* In the gen6 PRM Volume 1 Part 1: Graphics Core, Section 7.18.3.7.1
* (Surface Arrays For all surfaces other than separate stencil buffer):
*
* "[DevSNB] Errata: Sampler MSAA Qpitch will be 4 greater than the value
* calculated in the equation above , for every other odd Surface Height
* starting from 1 i.e. 1,5,9,13"
*
* Since this Qpitch errata only impacts the sampler, we have to adjust the
* input for the rendering surface to achieve the same qpitch. For the
* affected heights, we increment the height by 1 for the rendering
* surface.
*/
int height0 = irb->mt->logical_height0;
if (brw->gen == 6 && irb->mt->num_samples > 1 && (height0 % 4) == 1)
height0++;
surf[2] = SET_FIELD(mt->logical_width0 - 1, BRW_SURFACE_WIDTH) |
SET_FIELD(height0 - 1, BRW_SURFACE_HEIGHT) |
SET_FIELD(irb->mt_level - irb->mt->first_level, BRW_SURFACE_LOD);
surf[3] = brw_get_surface_tiling_bits(mt->tiling) |
SET_FIELD(depth - 1, BRW_SURFACE_DEPTH) |
SET_FIELD(mt->pitch - 1, BRW_SURFACE_PITCH);
surf[4] = brw_get_surface_num_multisamples(mt->num_samples) |
SET_FIELD(min_array_element, BRW_SURFACE_MIN_ARRAY_ELEMENT) |
SET_FIELD(depth - 1, BRW_SURFACE_RENDER_TARGET_VIEW_EXTENT);
surf[5] = (mt->align_h == 4 ? BRW_SURFACE_VERTICAL_ALIGN_ENABLE : 0);
drm_intel_bo_emit_reloc(brw->batch.bo,
brw->wm.base.surf_offset[surf_index] + 4,
mt->bo,
surf[1] - mt->bo->offset64,
I915_GEM_DOMAIN_RENDER,
I915_GEM_DOMAIN_RENDER);
}
/**
* Do additional "completeness" testing of a framebuffer object.
*/
static void
intel_validate_framebuffer(struct gl_context *ctx, struct gl_framebuffer *fb)
{
struct intel_context *intel = intel_context(ctx);
struct intel_renderbuffer *depthRb =
intel_get_renderbuffer(fb, BUFFER_DEPTH);
struct intel_renderbuffer *stencilRb =
intel_get_renderbuffer(fb, BUFFER_STENCIL);
struct intel_mipmap_tree *depth_mt = NULL, *stencil_mt = NULL;
int i;
DBG("%s() on fb %p (%s)\n", __FUNCTION__,
fb, (fb == ctx->DrawBuffer ? "drawbuffer" :
(fb == ctx->ReadBuffer ? "readbuffer" : "other buffer")));
if (depthRb)
depth_mt = depthRb->mt;
if (stencilRb)
stencil_mt = stencilRb->mt;
if (depth_mt && stencil_mt) {
/* Make sure that the depth and stencil buffers are actually the same
* slice of the same miptree, since we only support packed
* depth/stencil.
*/
if (depth_mt == stencil_mt) {
if (depthRb->mt_level != stencilRb->mt_level ||
depthRb->mt_layer != stencilRb->mt_layer) {
fbo_incomplete(fb,
"FBO incomplete: depth image level/layer %d/%d != "
"stencil image %d/%d\n",
depthRb->mt_level,
depthRb->mt_layer,
stencilRb->mt_level,
stencilRb->mt_layer);
}
} else {
fbo_incomplete(fb, "FBO incomplete: separate stencil unsupported\n");
}
}
for (i = 0; i < Elements(fb->Attachment); i++) {
struct gl_renderbuffer *rb;
struct intel_renderbuffer *irb;
if (fb->Attachment[i].Type == GL_NONE)
continue;
/* A supported attachment will have a Renderbuffer set either
* from being a Renderbuffer or being a texture that got the
* intel_wrap_texture() treatment.
*/
rb = fb->Attachment[i].Renderbuffer;
if (rb == NULL) {
fbo_incomplete(fb, "FBO incomplete: attachment without "
"renderbuffer\n");
continue;
}
if (fb->Attachment[i].Type == GL_TEXTURE) {
if (rb->TexImage->Border) {
fbo_incomplete(fb, "FBO incomplete: texture with border\n");
continue;
}
}
irb = intel_renderbuffer(rb);
if (irb == NULL) {
fbo_incomplete(fb, "FBO incomplete: software rendering "
"renderbuffer\n");
continue;
}
if (!intel->vtbl.render_target_supported(intel, rb)) {
fbo_incomplete(fb, "FBO incomplete: Unsupported HW "
"texture/renderbuffer format attached: %s\n",
_mesa_get_format_name(intel_rb_format(irb)));
}
}
}
/**
* Do additional "completeness" testing of a framebuffer object.
*/
static void
intel_validate_framebuffer(struct gl_context *ctx, struct gl_framebuffer *fb)
{
struct brw_context *brw = brw_context(ctx);
struct intel_renderbuffer *depthRb =
intel_get_renderbuffer(fb, BUFFER_DEPTH);
struct intel_renderbuffer *stencilRb =
intel_get_renderbuffer(fb, BUFFER_STENCIL);
struct intel_mipmap_tree *depth_mt = NULL, *stencil_mt = NULL;
int i;
DBG("%s() on fb %p (%s)\n", __FUNCTION__,
fb, (fb == ctx->DrawBuffer ? "drawbuffer" :
(fb == ctx->ReadBuffer ? "readbuffer" : "other buffer")));
if (depthRb)
depth_mt = depthRb->mt;
if (stencilRb) {
stencil_mt = stencilRb->mt;
if (stencil_mt->stencil_mt)
stencil_mt = stencil_mt->stencil_mt;
}
if (depth_mt && stencil_mt) {
if (brw->gen >= 7) {
/* For gen >= 7, we are using the lod/minimum-array-element fields
* and supportting layered rendering. This means that we must restrict
* the depth & stencil attachments to match in various more retrictive
* ways. (width, height, depth, LOD and layer)
*/
if (depth_mt->physical_width0 != stencil_mt->physical_width0 ||
depth_mt->physical_height0 != stencil_mt->physical_height0 ||
depth_mt->physical_depth0 != stencil_mt->physical_depth0 ||
depthRb->mt_level != stencilRb->mt_level ||
depthRb->mt_layer != stencilRb->mt_layer) {
fbo_incomplete(fb,
"FBO incomplete: depth and stencil must match in"
"width, height, depth, LOD and layer\n");
}
}
if (depth_mt == stencil_mt) {
/* For true packed depth/stencil (not faked on prefers-separate-stencil
* hardware) we need to be sure they're the same level/layer, since
* we'll be emitting a single packet describing the packed setup.
*/
if (depthRb->mt_level != stencilRb->mt_level ||
depthRb->mt_layer != stencilRb->mt_layer) {
fbo_incomplete(fb,
"FBO incomplete: depth image level/layer %d/%d != "
"stencil image %d/%d\n",
depthRb->mt_level,
depthRb->mt_layer,
stencilRb->mt_level,
stencilRb->mt_layer);
}
} else {
if (!brw->has_separate_stencil) {
fbo_incomplete(fb, "FBO incomplete: separate stencil "
"unsupported\n");
}
if (stencil_mt->format != MESA_FORMAT_S8) {
fbo_incomplete(fb, "FBO incomplete: separate stencil is %s "
"instead of S8\n",
_mesa_get_format_name(stencil_mt->format));
}
if (brw->gen < 7 && !intel_renderbuffer_has_hiz(depthRb)) {
/* Before Gen7, separate depth and stencil buffers can be used
* only if HiZ is enabled. From the Sandybridge PRM, Volume 2,
* Part 1, Bit 3DSTATE_DEPTH_BUFFER.SeparateStencilBufferEnable:
* [DevSNB]: This field must be set to the same value (enabled
* or disabled) as Hierarchical Depth Buffer Enable.
*/
fbo_incomplete(fb, "FBO incomplete: separate stencil "
"without HiZ\n");
}
}
}
for (i = 0; i < Elements(fb->Attachment); i++) {
struct gl_renderbuffer *rb;
struct intel_renderbuffer *irb;
if (fb->Attachment[i].Type == GL_NONE)
continue;
/* A supported attachment will have a Renderbuffer set either
* from being a Renderbuffer or being a texture that got the
* intel_wrap_texture() treatment.
*/
rb = fb->Attachment[i].Renderbuffer;
if (rb == NULL) {
fbo_incomplete(fb, "FBO incomplete: attachment without "
"renderbuffer\n");
continue;
}
if (fb->Attachment[i].Type == GL_TEXTURE) {
if (rb->TexImage->Border) {
fbo_incomplete(fb, "FBO incomplete: texture with border\n");
continue;
}
}
irb = intel_renderbuffer(rb);
if (irb == NULL) {
fbo_incomplete(fb, "FBO incomplete: software rendering "
"renderbuffer\n");
continue;
}
if (!brw_render_target_supported(brw, rb)) {
fbo_incomplete(fb, "FBO incomplete: Unsupported HW "
"texture/renderbuffer format attached: %s\n",
_mesa_get_format_name(intel_rb_format(irb)));
}
}
}
/*
* 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 brw_context *brw = brw_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(brw);
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 (_mesa_get_render_format(ctx, intel_rb_format(irb))) {
case MESA_FORMAT_ARGB8888:
case MESA_FORMAT_XRGB8888:
color = PACK_COLOR_8888(ubcolor[3], ubcolor[0], ubcolor[1], ubcolor[2]);
break;
case MESA_FORMAT_RGB565:
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
/* The blitter has no idea about fast color clears, so we need to resolve
* the miptree before we do anything.
*/
intel_miptree_resolve_color(brw, irb->mt);
/* 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(brw,
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(brw);
if (_mesa_is_bufferobj(unpack->BufferObj)) {
/* done with PBO so unmap it now */
ctx->Driver.UnmapBuffer(ctx, unpack->BufferObj);
}
intel_check_front_buffer_rendering(brw);
return 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_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 bool
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;
bool flip = false;
struct intel_renderbuffer *draw_irb = NULL;
struct intel_renderbuffer *read_irb = NULL;
gl_format read_format, draw_format;
/* Update draw buffer bounds */
_mesa_update_state(ctx);
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;
}
read_format = intel_rb_format(read_irb);
draw_format = intel_rb_format(draw_irb);
if (draw_format != read_format &&
!(draw_format == MESA_FORMAT_XRGB8888 &&
read_format == MESA_FORMAT_ARGB8888)) {
perf_debug("glCopyPixels() fallback: mismatched formats (%s -> %s\n",
_mesa_get_format_name(read_format),
_mesa_get_format_name(draw_format));
return 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 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 (_mesa_is_winsys_fbo(fb)) {
/* copypixels to a window system framebuffer */
dsty = fb->Height - dsty - height;
flip = !flip;
}
/* Flip source Y if it's a window system framebuffer. */
if (_mesa_is_winsys_fbo(read_fb)) {
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;
uint32_t src_pitch = read_irb->mt->region->pitch;
if (flip)
src_pitch = -src_pitch;
if (!intelEmitCopyBlit(intel,
draw_irb->mt->cpp,
src_pitch, read_irb->mt->region->bo,
0, read_irb->mt->region->tiling,
draw_irb->mt->region->pitch, draw_irb->mt->region->bo,
0, draw_irb->mt->region->tiling,
srcx, srcy,
dstx, dsty,
width, height,
ctx->Color.ColorLogicOpEnabled ?
ctx->Color.LogicOp : GL_COPY)) {
DBG("%s: blit failure\n", __FUNCTION__);
return false;
}
out:
intel_check_front_buffer_rendering(intel);
DBG("%s: success\n", __FUNCTION__);
return true;
}
