/**
* Polygon stipple offset packet
*/
static void
upload_polygon_stipple_offset(struct brw_context *brw)
{
struct gl_context *ctx = &brw->ctx;
/* _NEW_POLYGON */
if (!ctx->Polygon.StippleFlag)
return;
BEGIN_BATCH(2);
OUT_BATCH(_3DSTATE_POLY_STIPPLE_OFFSET << 16 | (2-2));
/* _NEW_BUFFERS
*
* If we're drawing to a system window we have to invert the Y axis
* in order to match the OpenGL pixel coordinate system, and our
* offset must be matched to the window position. If we're drawing
* to a user-created FBO then our native pixel coordinate system
* works just fine, and there's no window system to worry about.
*/
if (_mesa_is_winsys_fbo(ctx->DrawBuffer))
OUT_BATCH((32 - (_mesa_geometric_height(ctx->DrawBuffer) & 31)) & 31);
else
OUT_BATCH(0);
ADVANCE_BATCH();
}
/**
* Scissor depends on the scissor box, and the framebuffer dimensions.
*/
static void
update_scissor( struct st_context *st )
{
struct pipe_scissor_state scissor[PIPE_MAX_VIEWPORTS];
const struct gl_context *ctx = st->ctx;
const struct gl_framebuffer *fb = ctx->DrawBuffer;
const unsigned int fb_width = _mesa_geometric_width(fb);
const unsigned int fb_height = _mesa_geometric_height(fb);
GLint miny, maxy;
unsigned i;
bool changed = false;
for (i = 0 ; i < ctx->Const.MaxViewports; i++) {
scissor[i].minx = 0;
scissor[i].miny = 0;
scissor[i].maxx = fb_width;
scissor[i].maxy = fb_height;
if (ctx->Scissor.EnableFlags & (1 << i)) {
/* need to be careful here with xmax or ymax < 0 */
GLint xmax = MAX2(0, ctx->Scissor.ScissorArray[i].X + ctx->Scissor.ScissorArray[i].Width);
GLint ymax = MAX2(0, ctx->Scissor.ScissorArray[i].Y + ctx->Scissor.ScissorArray[i].Height);
if (ctx->Scissor.ScissorArray[i].X > (GLint)scissor[i].minx)
scissor[i].minx = ctx->Scissor.ScissorArray[i].X;
if (ctx->Scissor.ScissorArray[i].Y > (GLint)scissor[i].miny)
scissor[i].miny = ctx->Scissor.ScissorArray[i].Y;
if (xmax < (GLint) scissor[i].maxx)
scissor[i].maxx = xmax;
if (ymax < (GLint) scissor[i].maxy)
scissor[i].maxy = ymax;
/* check for null space */
if (scissor[i].minx >= scissor[i].maxx || scissor[i].miny >= scissor[i].maxy)
scissor[i].minx = scissor[i].miny = scissor[i].maxx = scissor[i].maxy = 0;
}
/* Now invert Y if needed.
* Gallium drivers use the convention Y=0=top for surfaces.
*/
if (st_fb_orientation(fb) == Y_0_TOP) {
miny = fb->Height - scissor[i].maxy;
maxy = fb->Height - scissor[i].miny;
scissor[i].miny = miny;
scissor[i].maxy = maxy;
}
if (memcmp(&scissor[i], &st->state.scissor[i], sizeof(scissor[0])) != 0) {
/* state has changed */
st->state.scissor[i] = scissor[i]; /* struct copy */
changed = true;
}
}
if (changed)
st->pipe->set_scissor_states(st->pipe, 0, ctx->Const.MaxViewports, scissor); /* activate */
}
/* Constant single cliprect for framebuffer object or DRI2 drawing */
static void
upload_drawing_rect(struct brw_context *brw)
{
struct gl_context *ctx = &brw->ctx;
const struct gl_framebuffer *fb = ctx->DrawBuffer;
const unsigned int fb_width = _mesa_geometric_width(fb);
const unsigned int fb_height = _mesa_geometric_height(fb);
BEGIN_BATCH(4);
OUT_BATCH(_3DSTATE_DRAWING_RECTANGLE << 16 | (4 - 2));
OUT_BATCH(0); /* xmin, ymin */
OUT_BATCH(((fb_width - 1) & 0xffff) | ((fb_height - 1) << 16));
OUT_BATCH(0);
ADVANCE_BATCH();
}
static void
gen6_upload_sf_vp(struct brw_context *brw)
{
struct gl_context *ctx = &brw->ctx;
struct gen6_sf_viewport *sfv;
GLfloat y_scale, y_bias;
const bool render_to_fbo = _mesa_is_user_fbo(ctx->DrawBuffer);
sfv = brw_state_batch(brw, AUB_TRACE_SF_VP_STATE,
sizeof(*sfv) * ctx->Const.MaxViewports,
32, &brw->sf.vp_offset);
memset(sfv, 0, sizeof(*sfv) * ctx->Const.MaxViewports);
/* _NEW_BUFFERS */
if (render_to_fbo) {
y_scale = 1.0;
y_bias = 0.0;
} else {
y_scale = -1.0;
y_bias = (float)_mesa_geometric_height(ctx->DrawBuffer);
}
for (unsigned i = 0; i < ctx->Const.MaxViewports; i++) {
float scale[3], translate[3];
/* _NEW_VIEWPORT */
_mesa_get_viewport_xform(ctx, i, scale, translate);
sfv[i].m00 = scale[0];
sfv[i].m11 = scale[1] * y_scale;
sfv[i].m22 = scale[2];
sfv[i].m30 = translate[0];
sfv[i].m31 = translate[1] * y_scale + y_bias;
sfv[i].m32 = translate[2];
}
brw->ctx.NewDriverState |= BRW_NEW_SF_VP;
}
static void
gen7_upload_sf_clip_viewport(struct brw_context *brw)
{
struct gl_context *ctx = &brw->ctx;
GLfloat y_scale, y_bias;
const bool render_to_fbo = _mesa_is_user_fbo(ctx->DrawBuffer);
struct gen7_sf_clip_viewport *vp;
/* BRW_NEW_VIEWPORT_COUNT */
const unsigned viewport_count = brw->clip.viewport_count;
vp = brw_state_batch(brw, AUB_TRACE_SF_VP_STATE,
sizeof(*vp) * viewport_count, 64,
&brw->sf.vp_offset);
/* Also assign to clip.vp_offset in case something uses it. */
brw->clip.vp_offset = brw->sf.vp_offset;
/* _NEW_BUFFERS */
if (render_to_fbo) {
y_scale = 1.0;
y_bias = 0.0;
} else {
y_scale = -1.0;
y_bias = (float)_mesa_geometric_height(ctx->DrawBuffer);
}
for (unsigned i = 0; i < viewport_count; i++) {
float scale[3], translate[3];
_mesa_get_viewport_xform(ctx, i, scale, translate);
/* According to the "Vertex X,Y Clamping and Quantization" section of
* the Strips and Fans documentation, objects must not have a
* screen-space extents of over 8192 pixels, or they may be
* mis-rasterized. The maximum screen space coordinates of a small
* object may larger, but we have no way to enforce the object size
* other than through clipping.
*
* If you're surprised that we set clip to -gbx to +gbx and it seems
* like we'll end up with 16384 wide, note that for a 8192-wide render
* target, we'll end up with a normal (-1, 1) clip volume that just
* covers the drawable.
*/
const float maximum_guardband_extent = 8192;
const float gbx = maximum_guardband_extent / ctx->ViewportArray[i].Width;
const float gby = maximum_guardband_extent / ctx->ViewportArray[i].Height;
vp[i].guardband.xmin = -gbx;
vp[i].guardband.xmax = gbx;
vp[i].guardband.ymin = -gby;
vp[i].guardband.ymax = gby;
/* _NEW_VIEWPORT */
vp[i].viewport.m00 = scale[0];
vp[i].viewport.m11 = scale[1] * y_scale;
vp[i].viewport.m22 = scale[2];
vp[i].viewport.m30 = translate[0];
vp[i].viewport.m31 = translate[1] * y_scale + y_bias;
vp[i].viewport.m32 = translate[2];
}
BEGIN_BATCH(2);
OUT_BATCH(_3DSTATE_VIEWPORT_STATE_POINTERS_SF_CL << 16 | (2 - 2));
OUT_BATCH(brw->sf.vp_offset);
ADVANCE_BATCH();
}
static void
brw_wm_populate_key(struct brw_context *brw, struct brw_wm_prog_key *key)
{
struct gl_context *ctx = &brw->ctx;
/* BRW_NEW_FRAGMENT_PROGRAM */
const struct brw_fragment_program *fp =
(struct brw_fragment_program *) brw->fragment_program;
const struct gl_program *prog = (struct gl_program *) brw->fragment_program;
GLuint lookup = 0;
GLuint line_aa;
bool program_uses_dfdy = fp->program.UsesDFdy;
const bool multisample_fbo = _mesa_geometric_samples(ctx->DrawBuffer) > 1;
memset(key, 0, sizeof(*key));
/* Build the index for table lookup
*/
if (brw->gen < 6) {
/* _NEW_COLOR */
if (fp->program.UsesKill || ctx->Color.AlphaEnabled)
lookup |= IZ_PS_KILL_ALPHATEST_BIT;
if (fp->program.Base.OutputsWritten & BITFIELD64_BIT(FRAG_RESULT_DEPTH))
lookup |= IZ_PS_COMPUTES_DEPTH_BIT;
/* _NEW_DEPTH */
if (ctx->Depth.Test)
lookup |= IZ_DEPTH_TEST_ENABLE_BIT;
if (ctx->Depth.Test && ctx->Depth.Mask) /* ?? */
lookup |= IZ_DEPTH_WRITE_ENABLE_BIT;
/* _NEW_STENCIL | _NEW_BUFFERS */
if (ctx->Stencil._Enabled) {
lookup |= IZ_STENCIL_TEST_ENABLE_BIT;
if (ctx->Stencil.WriteMask[0] ||
ctx->Stencil.WriteMask[ctx->Stencil._BackFace])
lookup |= IZ_STENCIL_WRITE_ENABLE_BIT;
}
key->iz_lookup = lookup;
}
line_aa = AA_NEVER;
/* _NEW_LINE, _NEW_POLYGON, BRW_NEW_REDUCED_PRIMITIVE */
if (ctx->Line.SmoothFlag) {
if (brw->reduced_primitive == GL_LINES) {
line_aa = AA_ALWAYS;
}
else if (brw->reduced_primitive == GL_TRIANGLES) {
if (ctx->Polygon.FrontMode == GL_LINE) {
line_aa = AA_SOMETIMES;
if (ctx->Polygon.BackMode == GL_LINE ||
(ctx->Polygon.CullFlag &&
ctx->Polygon.CullFaceMode == GL_BACK))
line_aa = AA_ALWAYS;
}
else if (ctx->Polygon.BackMode == GL_LINE) {
line_aa = AA_SOMETIMES;
if ((ctx->Polygon.CullFlag &&
ctx->Polygon.CullFaceMode == GL_FRONT))
line_aa = AA_ALWAYS;
}
}
}
key->line_aa = line_aa;
/* _NEW_HINT */
key->high_quality_derivatives =
ctx->Hint.FragmentShaderDerivative == GL_NICEST;
if (brw->gen < 6)
key->stats_wm = brw->stats_wm;
/* _NEW_LIGHT */
key->flat_shade = (ctx->Light.ShadeModel == GL_FLAT);
/* _NEW_FRAG_CLAMP | _NEW_BUFFERS */
key->clamp_fragment_color = ctx->Color._ClampFragmentColor;
/* _NEW_TEXTURE */
brw_populate_sampler_prog_key_data(ctx, prog, brw->wm.base.sampler_count,
&key->tex);
/* _NEW_BUFFERS */
/*
* Include the draw buffer origin and height so that we can calculate
* fragment position values relative to the bottom left of the drawable,
* from the incoming screen origin relative position we get as part of our
* payload.
*
* This is only needed for the WM_WPOSXY opcode when the fragment program
* uses the gl_FragCoord input.
*
* We could avoid recompiling by including this as a constant referenced by
* our program, but if we were to do that it would also be nice to handle
* getting that constant updated at batchbuffer submit time (when we
* hold the lock and know where the buffer really is) rather than at emit
* time when we don't hold the lock and are just guessing. We could also
* just avoid using this as key data if the program doesn't use
* fragment.position.
*
* For DRI2 the origin_x/y will always be (0,0) but we still need the
* drawable height in order to invert the Y axis.
*/
if (fp->program.Base.InputsRead & VARYING_BIT_POS) {
key->drawable_height = _mesa_geometric_height(ctx->DrawBuffer);
}
if ((fp->program.Base.InputsRead & VARYING_BIT_POS) || program_uses_dfdy) {
key->render_to_fbo = _mesa_is_user_fbo(ctx->DrawBuffer);
}
/* _NEW_BUFFERS */
key->nr_color_regions = ctx->DrawBuffer->_NumColorDrawBuffers;
/* _NEW_MULTISAMPLE, _NEW_COLOR, _NEW_BUFFERS */
key->replicate_alpha = ctx->DrawBuffer->_NumColorDrawBuffers > 1 &&
(ctx->Multisample.SampleAlphaToCoverage || ctx->Color.AlphaEnabled);
/* _NEW_BUFFERS _NEW_MULTISAMPLE */
/* Ignore sample qualifier while computing this flag. */
key->persample_shading =
_mesa_get_min_invocations_per_fragment(ctx, &fp->program, true) > 1;
if (key->persample_shading)
key->persample_2x = _mesa_geometric_samples(ctx->DrawBuffer) == 2;
key->compute_pos_offset =
_mesa_get_min_invocations_per_fragment(ctx, &fp->program, false) > 1 &&
fp->program.Base.SystemValuesRead & SYSTEM_BIT_SAMPLE_POS;
key->compute_sample_id =
multisample_fbo &&
ctx->Multisample.Enabled &&
(fp->program.Base.SystemValuesRead & SYSTEM_BIT_SAMPLE_ID);
/* BRW_NEW_VUE_MAP_GEOM_OUT */
if (brw->gen < 6 || _mesa_bitcount_64(fp->program.Base.InputsRead &
BRW_FS_VARYING_INPUT_MASK) > 16)
key->input_slots_valid = brw->vue_map_geom_out.slots_valid;
/* _NEW_COLOR | _NEW_BUFFERS */
/* Pre-gen6, the hardware alpha test always used each render
* target's alpha to do alpha test, as opposed to render target 0's alpha
* like GL requires. Fix that by building the alpha test into the
* shader, and we'll skip enabling the fixed function alpha test.
*/
if (brw->gen < 6 && ctx->DrawBuffer->_NumColorDrawBuffers > 1 &&
ctx->Color.AlphaEnabled) {
key->alpha_test_func = ctx->Color.AlphaFunc;
key->alpha_test_ref = ctx->Color.AlphaRef;
}
/* The unique fragment program ID */
key->program_string_id = fp->id;
}
static void
upload_clip_state(struct brw_context *brw)
{
struct gl_context *ctx = &brw->ctx;
/* BRW_NEW_META_IN_PROGRESS */
uint32_t dw1 = brw->meta_in_progress ? 0 : GEN6_CLIP_STATISTICS_ENABLE;
uint32_t dw2 = 0;
/* _NEW_BUFFERS */
struct gl_framebuffer *fb = ctx->DrawBuffer;
/* BRW_NEW_FS_PROG_DATA */
if (brw->wm.prog_data->barycentric_interp_modes &
BRW_WM_NONPERSPECTIVE_BARYCENTRIC_BITS) {
dw2 |= GEN6_CLIP_NON_PERSPECTIVE_BARYCENTRIC_ENABLE;
}
dw1 |= brw->vs.prog_data->base.cull_distance_mask;
if (brw->gen >= 7)
dw1 |= GEN7_CLIP_EARLY_CULL;
if (brw->gen == 7) {
/* _NEW_POLYGON */
if (ctx->Polygon._FrontBit == _mesa_is_user_fbo(fb))
dw1 |= GEN7_CLIP_WINDING_CCW;
if (ctx->Polygon.CullFlag) {
switch (ctx->Polygon.CullFaceMode) {
case GL_FRONT:
dw1 |= GEN7_CLIP_CULLMODE_FRONT;
break;
case GL_BACK:
dw1 |= GEN7_CLIP_CULLMODE_BACK;
break;
case GL_FRONT_AND_BACK:
dw1 |= GEN7_CLIP_CULLMODE_BOTH;
break;
default:
unreachable("Should not get here: invalid CullFlag");
}
} else {
dw1 |= GEN7_CLIP_CULLMODE_NONE;
}
}
if (brw->gen < 8 && !ctx->Transform.DepthClamp)
dw2 |= GEN6_CLIP_Z_TEST;
/* _NEW_LIGHT */
if (ctx->Light.ProvokingVertex == GL_FIRST_VERTEX_CONVENTION) {
dw2 |=
(0 << GEN6_CLIP_TRI_PROVOKE_SHIFT) |
(1 << GEN6_CLIP_TRIFAN_PROVOKE_SHIFT) |
(0 << GEN6_CLIP_LINE_PROVOKE_SHIFT);
} else {
dw2 |=
(2 << GEN6_CLIP_TRI_PROVOKE_SHIFT) |
(2 << GEN6_CLIP_TRIFAN_PROVOKE_SHIFT) |
(1 << GEN6_CLIP_LINE_PROVOKE_SHIFT);
}
/* _NEW_TRANSFORM */
dw2 |= (ctx->Transform.ClipPlanesEnabled <<
GEN6_USER_CLIP_CLIP_DISTANCES_SHIFT);
if (ctx->Transform.ClipDepthMode == GL_ZERO_TO_ONE)
dw2 |= GEN6_CLIP_API_D3D;
else
dw2 |= GEN6_CLIP_API_OGL;
dw2 |= GEN6_CLIP_GB_TEST;
/* We need to disable guardband clipping if the guardband (which we always
* program to the maximum screen-space bounding box of 8K x 8K) will be
* smaller than the viewport.
*
* Closely examining the clip determination formulas in the documentation
* reveals that objects will be discarded entirely if they're outside the
* (small) guardband, even if they're within the (large) viewport:
*
* TR = TR_GB || TR_VPXY || TR_VPZ || TR_UC || TR_NEGW
* TA = !TR && TA_GB && TA_VPZ && TA_NEGW
* MC = !(TA || TR)
*
* (TA is "Trivial Accept", TR is "Trivial Reject", MC is "Must Clip".)
*
* Disabling guardband clipping removes the TR_GB condition, which means
* they'll be considered MC ("Must Clip") unless they're rejected for
* some other reason.
*
* Note that there is no TA_VPXY condition. If there were, objects entirely
* inside a 16384x16384 viewport would be trivially accepted, breaking the
* "objects must have a screenspace bounding box not exceeding 8K in the X
* or Y direction" restriction. Instead, they're clipped.
*/
for (unsigned i = 0; i < ctx->Const.MaxViewports; i++) {
if (ctx->ViewportArray[i].Width > 8192 ||
ctx->ViewportArray[i].Height > 8192) {
dw2 &= ~GEN6_CLIP_GB_TEST;
break;
}
}
/* If the viewport dimensions are smaller than the drawable dimensions,
* we have to disable guardband clipping prior to Gen8. We always program
* the guardband to a fixed size, which is almost always larger than the
* viewport. Any geometry which intersects the viewport but lies within
* the guardband would bypass the 3D clipping stage, so it wouldn't be
* clipped to the viewport. Rendering would happen beyond the viewport,
* but still inside the drawable.
*
* Gen8+ introduces a viewport extents test which restricts rendering to
* the viewport, so we can ignore this restriction.
*/
if (brw->gen < 8) {
const float fb_width = (float)_mesa_geometric_width(fb);
const float fb_height = (float)_mesa_geometric_height(fb);
for (unsigned i = 0; i < ctx->Const.MaxViewports; i++) {
if (ctx->ViewportArray[i].X != 0 ||
ctx->ViewportArray[i].Y != 0 ||
ctx->ViewportArray[i].Width != fb_width ||
ctx->ViewportArray[i].Height != fb_height) {
dw2 &= ~GEN6_CLIP_GB_TEST;
break;
}
}
}
/* BRW_NEW_RASTERIZER_DISCARD */
if (ctx->RasterDiscard) {
dw2 |= GEN6_CLIP_MODE_REJECT_ALL;
if (brw->gen == 6) {
perf_debug("Rasterizer discard is currently implemented via the "
"clipper; having the GS not write primitives would "
"likely be faster.\n");
}
}
uint32_t enable;
if (brw->primitive == _3DPRIM_RECTLIST)
enable = 0;
else
enable = GEN6_CLIP_ENABLE;
if (!is_drawing_points(brw) && !is_drawing_lines(brw))
dw2 |= GEN6_CLIP_XY_TEST;
/* BRW_NEW_VUE_MAP_GEOM_OUT */
const int max_vp_index =
(brw->vue_map_geom_out.slots_valid & VARYING_BIT_VIEWPORT) != 0 ?
ctx->Const.MaxViewports : 1;
BEGIN_BATCH(4);
OUT_BATCH(_3DSTATE_CLIP << 16 | (4 - 2));
OUT_BATCH(dw1);
OUT_BATCH(enable |
GEN6_CLIP_MODE_NORMAL |
dw2);
OUT_BATCH(U_FIXED(0.125, 3) << GEN6_CLIP_MIN_POINT_WIDTH_SHIFT |
U_FIXED(255.875, 3) << GEN6_CLIP_MAX_POINT_WIDTH_SHIFT |
(_mesa_geometric_layers(fb) > 0 ? 0 : GEN6_CLIP_FORCE_ZERO_RTAINDEX) |
((max_vp_index - 1) & GEN6_CLIP_MAX_VP_INDEX_MASK));
ADVANCE_BATCH();
}
/**
* Update framebuffer state (color, depth, stencil, etc. buffers)
*/
void
st_update_framebuffer_state( struct st_context *st )
{
struct pipe_framebuffer_state framebuffer;
struct gl_framebuffer *fb = st->ctx->DrawBuffer;
struct st_renderbuffer *strb;
GLuint i;
st_flush_bitmap_cache(st);
st_invalidate_readpix_cache(st);
st->state.fb_orientation = st_fb_orientation(fb);
/**
* Quantize the derived default number of samples:
*
* A query to the driver of supported MSAA values the
* hardware supports is done as to legalize the number
* of application requested samples, NumSamples.
* See commit eb9cf3c for more information.
*/
fb->DefaultGeometry._NumSamples =
framebuffer_quantize_num_samples(st, fb->DefaultGeometry.NumSamples);
framebuffer.width = _mesa_geometric_width(fb);
framebuffer.height = _mesa_geometric_height(fb);
framebuffer.samples = _mesa_geometric_samples(fb);
framebuffer.layers = _mesa_geometric_layers(fb);
/* Examine Mesa's ctx->DrawBuffer->_ColorDrawBuffers state
* to determine which surfaces to draw to
*/
framebuffer.nr_cbufs = fb->_NumColorDrawBuffers;
for (i = 0; i < fb->_NumColorDrawBuffers; i++) {
framebuffer.cbufs[i] = NULL;
strb = st_renderbuffer(fb->_ColorDrawBuffers[i]);
if (strb) {
if (strb->is_rtt || (strb->texture &&
_mesa_get_format_color_encoding(strb->Base.Format) == GL_SRGB)) {
/* rendering to a GL texture, may have to update surface */
st_update_renderbuffer_surface(st, strb);
}
if (strb->surface) {
framebuffer.cbufs[i] = strb->surface;
update_framebuffer_size(&framebuffer, strb->surface);
}
strb->defined = GL_TRUE; /* we'll be drawing something */
}
}
for (i = framebuffer.nr_cbufs; i < PIPE_MAX_COLOR_BUFS; i++) {
framebuffer.cbufs[i] = NULL;
}
/* Remove trailing GL_NONE draw buffers. */
while (framebuffer.nr_cbufs &&
!framebuffer.cbufs[framebuffer.nr_cbufs-1]) {
framebuffer.nr_cbufs--;
}
/*
* Depth/Stencil renderbuffer/surface.
*/
strb = st_renderbuffer(fb->Attachment[BUFFER_DEPTH].Renderbuffer);
if (!strb)
strb = st_renderbuffer(fb->Attachment[BUFFER_STENCIL].Renderbuffer);
if (strb) {
if (strb->is_rtt) {
/* rendering to a GL texture, may have to update surface */
st_update_renderbuffer_surface(st, strb);
}
framebuffer.zsbuf = strb->surface;
update_framebuffer_size(&framebuffer, strb->surface);
}
else
framebuffer.zsbuf = NULL;
#ifdef DEBUG
/* Make sure the resource binding flags were set properly */
for (i = 0; i < framebuffer.nr_cbufs; i++) {
assert(!framebuffer.cbufs[i] ||
framebuffer.cbufs[i]->texture->bind & PIPE_BIND_RENDER_TARGET);
}
if (framebuffer.zsbuf) {
assert(framebuffer.zsbuf->texture->bind & PIPE_BIND_DEPTH_STENCIL);
}
#endif
if (framebuffer.width == USHRT_MAX)
framebuffer.width = 0;
if (framebuffer.height == USHRT_MAX)
framebuffer.height = 0;
cso_set_framebuffer(st->cso_context, &framebuffer);
st->state.fb_width = framebuffer.width;
st->state.fb_height = framebuffer.height;
st->state.fb_num_samples = util_framebuffer_get_num_samples(&framebuffer);
st->state.fb_num_layers = util_framebuffer_get_num_layers(&framebuffer);
st->state.fb_num_cb = framebuffer.nr_cbufs;
}
static void
st_DrawTex(struct gl_context *ctx, GLfloat x, GLfloat y, GLfloat z,
GLfloat width, GLfloat height)
{
struct st_context *st = ctx->st;
struct pipe_context *pipe = st->pipe;
struct cso_context *cso = st->cso_context;
struct pipe_resource *vbuffer = NULL;
GLuint i, numTexCoords, numAttribs;
GLboolean emitColor;
uint semantic_names[2 + MAX_TEXTURE_UNITS];
uint semantic_indexes[2 + MAX_TEXTURE_UNITS];
struct pipe_vertex_element velements[2 + MAX_TEXTURE_UNITS];
unsigned offset;
st_flush_bitmap_cache(st);
st_invalidate_readpix_cache(st);
st_validate_state(st, ST_PIPELINE_RENDER);
/* determine if we need vertex color */
if (ctx->FragmentProgram._Current->info.inputs_read & VARYING_BIT_COL0)
emitColor = GL_TRUE;
else
emitColor = GL_FALSE;
/* determine how many enabled sets of texcoords */
numTexCoords = 0;
for (i = 0; i < ctx->Const.MaxTextureUnits; i++) {
if (ctx->Texture.Unit[i]._Current &&
ctx->Texture.Unit[i]._Current->Target == GL_TEXTURE_2D) {
numTexCoords++;
}
}
/* total number of attributes per vertex */
numAttribs = 1 + emitColor + numTexCoords;
/* load vertex buffer */
{
#define SET_ATTRIB(VERT, ATTR, X, Y, Z, W) \
do { \
GLuint k = (((VERT) * numAttribs + (ATTR)) * 4); \
assert(k < 4 * 4 * numAttribs); \
vbuf[k + 0] = X; \
vbuf[k + 1] = Y; \
vbuf[k + 2] = Z; \
vbuf[k + 3] = W; \
} while (0)
const GLfloat x0 = x, y0 = y, x1 = x + width, y1 = y + height;
GLfloat *vbuf = NULL;
GLuint tex_attr;
u_upload_alloc(pipe->stream_uploader, 0,
numAttribs * 4 * 4 * sizeof(GLfloat), 4,
&offset, &vbuffer, (void **) &vbuf);
if (!vbuffer) {
return;
}
z = CLAMP(z, 0.0f, 1.0f);
/* positions (in clip coords) */
{
const struct gl_framebuffer *fb = ctx->DrawBuffer;
const GLfloat fb_width = (GLfloat)_mesa_geometric_width(fb);
const GLfloat fb_height = (GLfloat)_mesa_geometric_height(fb);
const GLfloat clip_x0 = (GLfloat)(x0 / fb_width * 2.0 - 1.0);
const GLfloat clip_y0 = (GLfloat)(y0 / fb_height * 2.0 - 1.0);
const GLfloat clip_x1 = (GLfloat)(x1 / fb_width * 2.0 - 1.0);
const GLfloat clip_y1 = (GLfloat)(y1 / fb_height * 2.0 - 1.0);
SET_ATTRIB(0, 0, clip_x0, clip_y0, z, 1.0f); /* lower left */
SET_ATTRIB(1, 0, clip_x1, clip_y0, z, 1.0f); /* lower right */
SET_ATTRIB(2, 0, clip_x1, clip_y1, z, 1.0f); /* upper right */
SET_ATTRIB(3, 0, clip_x0, clip_y1, z, 1.0f); /* upper left */
semantic_names[0] = TGSI_SEMANTIC_POSITION;
semantic_indexes[0] = 0;
}
/* colors */
if (emitColor) {
const GLfloat *c = ctx->Current.Attrib[VERT_ATTRIB_COLOR0];
SET_ATTRIB(0, 1, c[0], c[1], c[2], c[3]);
SET_ATTRIB(1, 1, c[0], c[1], c[2], c[3]);
SET_ATTRIB(2, 1, c[0], c[1], c[2], c[3]);
SET_ATTRIB(3, 1, c[0], c[1], c[2], c[3]);
semantic_names[1] = TGSI_SEMANTIC_COLOR;
semantic_indexes[1] = 0;
tex_attr = 2;
}
else {
tex_attr = 1;
}
/* texcoords */
for (i = 0; i < ctx->Const.MaxTextureUnits; i++) {
if (ctx->Texture.Unit[i]._Current &&
ctx->Texture.Unit[i]._Current->Target == GL_TEXTURE_2D) {
struct gl_texture_object *obj = ctx->Texture.Unit[i]._Current;
const struct gl_texture_image *img = _mesa_base_tex_image(obj);
const GLfloat wt = (GLfloat) img->Width;
const GLfloat ht = (GLfloat) img->Height;
const GLfloat s0 = obj->CropRect[0] / wt;
const GLfloat t0 = obj->CropRect[1] / ht;
const GLfloat s1 = (obj->CropRect[0] + obj->CropRect[2]) / wt;
const GLfloat t1 = (obj->CropRect[1] + obj->CropRect[3]) / ht;
/*printf("crop texcoords: %g, %g .. %g, %g\n", s0, t0, s1, t1);*/
SET_ATTRIB(0, tex_attr, s0, t0, 0.0f, 1.0f); /* lower left */
SET_ATTRIB(1, tex_attr, s1, t0, 0.0f, 1.0f); /* lower right */
SET_ATTRIB(2, tex_attr, s1, t1, 0.0f, 1.0f); /* upper right */
SET_ATTRIB(3, tex_attr, s0, t1, 0.0f, 1.0f); /* upper left */
semantic_names[tex_attr] = st->needs_texcoord_semantic ?
TGSI_SEMANTIC_TEXCOORD : TGSI_SEMANTIC_GENERIC;
/* XXX: should this use semantic index i instead of 0 ? */
semantic_indexes[tex_attr] = 0;
tex_attr++;
}
}
u_upload_unmap(pipe->stream_uploader);
#undef SET_ATTRIB
}
cso_save_state(cso, (CSO_BIT_VIEWPORT |
CSO_BIT_STREAM_OUTPUTS |
CSO_BIT_VERTEX_SHADER |
CSO_BIT_TESSCTRL_SHADER |
CSO_BIT_TESSEVAL_SHADER |
CSO_BIT_GEOMETRY_SHADER |
CSO_BIT_VERTEX_ELEMENTS |
CSO_BIT_AUX_VERTEX_BUFFER_SLOT));
{
void *vs = lookup_shader(pipe, numAttribs,
semantic_names, semantic_indexes);
cso_set_vertex_shader_handle(cso, vs);
}
cso_set_tessctrl_shader_handle(cso, NULL);
cso_set_tesseval_shader_handle(cso, NULL);
cso_set_geometry_shader_handle(cso, NULL);
for (i = 0; i < numAttribs; i++) {
velements[i].src_offset = i * 4 * sizeof(float);
velements[i].instance_divisor = 0;
velements[i].vertex_buffer_index = 0;
velements[i].src_format = PIPE_FORMAT_R32G32B32A32_FLOAT;
}
cso_set_vertex_elements(cso, numAttribs, velements);
cso_set_stream_outputs(cso, 0, NULL, NULL);
/* viewport state: viewport matching window dims */
{
const struct gl_framebuffer *fb = ctx->DrawBuffer;
const GLboolean invert = (st_fb_orientation(fb) == Y_0_TOP);
const GLfloat width = (GLfloat)_mesa_geometric_width(fb);
const GLfloat height = (GLfloat)_mesa_geometric_height(fb);
struct pipe_viewport_state vp;
vp.scale[0] = 0.5f * width;
vp.scale[1] = height * (invert ? -0.5f : 0.5f);
vp.scale[2] = 1.0f;
vp.translate[0] = 0.5f * width;
vp.translate[1] = 0.5f * height;
vp.translate[2] = 0.0f;
cso_set_viewport(cso, &vp);
}
util_draw_vertex_buffer(pipe, cso, vbuffer,
cso_get_aux_vertex_buffer_slot(cso),
offset, /* offset */
PIPE_PRIM_TRIANGLE_FAN,
4, /* verts */
numAttribs); /* attribs/vert */
pipe_resource_reference(&vbuffer, NULL);
/* restore state */
cso_restore_state(cso);
}
static void
brw_upload_clip_unit(struct brw_context *brw)
{
struct gl_context *ctx = &brw->ctx;
struct brw_clip_unit_state *clip;
/* _NEW_BUFFERS */
const struct gl_framebuffer *fb = ctx->DrawBuffer;
const float fb_width = (float)_mesa_geometric_width(fb);
const float fb_height = (float)_mesa_geometric_height(fb);
upload_clip_vp(brw);
clip = brw_state_batch(brw, AUB_TRACE_CLIP_STATE,
sizeof(*clip), 32, &brw->clip.state_offset);
memset(clip, 0, sizeof(*clip));
/* BRW_NEW_PROGRAM_CACHE | BRW_NEW_CLIP_PROG_DATA */
clip->thread0.grf_reg_count = (ALIGN(brw->clip.prog_data->total_grf, 16) /
16 - 1);
clip->thread0.kernel_start_pointer =
brw_program_reloc(brw,
brw->clip.state_offset +
offsetof(struct brw_clip_unit_state, thread0),
brw->clip.prog_offset +
(clip->thread0.grf_reg_count << 1)) >> 6;
clip->thread1.floating_point_mode = BRW_FLOATING_POINT_NON_IEEE_754;
clip->thread1.single_program_flow = 1;
clip->thread3.urb_entry_read_length = brw->clip.prog_data->urb_read_length;
clip->thread3.const_urb_entry_read_length =
brw->clip.prog_data->curb_read_length;
/* BRW_NEW_CURBE_OFFSETS */
clip->thread3.const_urb_entry_read_offset = brw->curbe.clip_start * 2;
clip->thread3.dispatch_grf_start_reg = 1;
clip->thread3.urb_entry_read_offset = 0;
/* BRW_NEW_URB_FENCE */
clip->thread4.nr_urb_entries = brw->urb.nr_clip_entries;
clip->thread4.urb_entry_allocation_size = brw->urb.vsize - 1;
/* If we have enough clip URB entries to run two threads, do so.
*/
if (brw->urb.nr_clip_entries >= 10) {
/* Half of the URB entries go to each thread, and it has to be an
* even number.
*/
assert(brw->urb.nr_clip_entries % 2 == 0);
/* Although up to 16 concurrent Clip threads are allowed on Ironlake,
* only 2 threads can output VUEs at a time.
*/
if (brw->gen == 5)
clip->thread4.max_threads = 16 - 1;
else
clip->thread4.max_threads = 2 - 1;
} else {
assert(brw->urb.nr_clip_entries >= 5);
clip->thread4.max_threads = 1 - 1;
}
if (unlikely(INTEL_DEBUG & DEBUG_STATS))
clip->thread4.stats_enable = 1;
/* _NEW_TRANSFORM */
if (brw->gen == 5 || brw->is_g4x)
clip->clip5.userclip_enable_flags = ctx->Transform.ClipPlanesEnabled;
else
/* Up to 6 actual clip flags, plus the 7th for negative RHW workaround. */
clip->clip5.userclip_enable_flags = (ctx->Transform.ClipPlanesEnabled & 0x3f) | 0x40;
clip->clip5.userclip_must_clip = 1;
/* enable guardband clipping if we can */
if (ctx->ViewportArray[0].X == 0 &&
ctx->ViewportArray[0].Y == 0 &&
ctx->ViewportArray[0].Width == fb_width &&
ctx->ViewportArray[0].Height == fb_height)
{
clip->clip5.guard_band_enable = 1;
clip->clip6.clipper_viewport_state_ptr =
(brw->batch.bo->offset64 + brw->clip.vp_offset) >> 5;
/* emit clip viewport relocation */
drm_intel_bo_emit_reloc(brw->batch.bo,
(brw->clip.state_offset +
offsetof(struct brw_clip_unit_state, clip6)),
brw->batch.bo, brw->clip.vp_offset,
I915_GEM_DOMAIN_INSTRUCTION, 0);
}
