void
nv10_emit_tex_gen(struct gl_context *ctx, int emit)
{
const int i = emit - NOUVEAU_STATE_TEX_GEN0;
struct nouveau_context *nctx = to_nouveau_context(ctx);
struct nouveau_pushbuf *push = context_push(ctx);
struct gl_texture_unit *unit = &ctx->Texture.Unit[i];
int j;
for (j = 0; j < 4; j++) {
if (nctx->fallback == HWTNL && (unit->TexGenEnabled & 1 << j)) {
struct gl_texgen *coord = get_texgen_coord(unit, j);
float *k = get_texgen_coeff(coord);
if (k) {
BEGIN_NV04(push, NV10_3D(TEX_GEN_COEFF(i, j)), 4);
PUSH_DATAp(push, k, 4);
}
BEGIN_NV04(push, NV10_3D(TEX_GEN_MODE(i,j)), 1);
PUSH_DATA (push, nvgl_texgen_mode(coord->Mode));
} else {
BEGIN_NV04(push, NV10_3D(TEX_GEN_MODE(i,j)), 1);
PUSH_DATA (push, 0);
}
}
context_dirty_i(ctx, TEX_MAT, i);
}
void
NV10EXAComposite(PixmapPtr pix_dst,
int srcX, int srcY,
int maskX, int maskY,
int dstX, int dstY,
int width, int height)
{
ScrnInfoPtr pScrn = xf86Screens[pix_dst->drawable.pScreen->myNum];
NVPtr pNv = NVPTR(pScrn);
struct nouveau_channel *chan = pNv->chan;
PicturePtr mask = pNv->pmpict,
src = pNv->pspict;
PictVector dstq[4] = QUAD(dstX, dstY, width, height),
maskq[4] = QUAD(maskX, maskY, width, height),
srcq[4] = QUAD(srcX, srcY, width, height);
MAP(transform_vertex, src->transform, srcq);
if (mask)
MAP(transform_vertex, mask->transform, maskq);
WAIT_RING (chan, 64);
BEGIN_NV04(chan, NV10_3D(VERTEX_BEGIN_END), 1);
OUT_RING (chan, NV10_3D_VERTEX_BEGIN_END_QUADS);
MAP(emit_vertex, pNv, dstq, srcq, mask ? maskq : NULL);
BEGIN_NV04(chan, NV10_3D(VERTEX_BEGIN_END), 1);
OUT_RING (chan, NV10_3D_VERTEX_BEGIN_END_STOP);
}
static void
setup_blend_function(NVPtr pNv, PicturePtr pdpict, int alu)
{
struct nouveau_pushbuf *push = pNv->pushbuf;
struct pict_op *op = &nv10_pict_op[alu];
int src_factor = op->src;
int dst_factor = op->dst;
if (src_factor == SF(ONE_MINUS_DST_ALPHA) &&
!PICT_FORMAT_A(pdpict->format))
/* ONE_MINUS_DST_ALPHA doesn't always do the right thing for
* framebuffers without alpha channel. But it's the same as
* ZERO in that case.
*/
src_factor = SF(ZERO);
if (effective_component_alpha(pNv->pmpict)) {
if (dst_factor == DF(SRC_ALPHA))
dst_factor = DF(SRC_COLOR);
else if (dst_factor == DF(ONE_MINUS_SRC_ALPHA))
dst_factor = DF(ONE_MINUS_SRC_COLOR);
}
BEGIN_NV04(push, NV10_3D(BLEND_FUNC_SRC), 2);
PUSH_DATA (push, src_factor);
PUSH_DATA (push, dst_factor);
BEGIN_NV04(push, NV10_3D(BLEND_FUNC_ENABLE), 1);
PUSH_DATA (push, 1);
}
void
nv10_emit_blend_equation(struct gl_context *ctx, int emit)
{
struct nouveau_pushbuf *push = context_push(ctx);
BEGIN_NV04(push, NV10_3D(BLEND_FUNC_ENABLE), 1);
PUSH_DATAb(push, ctx->Color.BlendEnabled);
BEGIN_NV04(push, NV10_3D(BLEND_EQUATION), 1);
PUSH_DATA (push, nvgl_blend_eqn(ctx->Color.Blend[0].EquationRGB));
}
void
nv10_emit_alpha_func(struct gl_context *ctx, int emit)
{
struct nouveau_pushbuf *push = context_push(ctx);
BEGIN_NV04(push, NV10_3D(ALPHA_FUNC_ENABLE), 1);
PUSH_DATAb(push, ctx->Color.AlphaEnabled);
BEGIN_NV04(push, NV10_3D(ALPHA_FUNC_FUNC), 2);
PUSH_DATA (push, nvgl_comparison_op(ctx->Color.AlphaFunc));
PUSH_DATA (push, FLOAT_TO_UBYTE(ctx->Color.AlphaRef));
}
void
nv10_emit_depth(struct gl_context *ctx, int emit)
{
struct nouveau_pushbuf *push = context_push(ctx);
BEGIN_NV04(push, NV10_3D(DEPTH_TEST_ENABLE), 1);
PUSH_DATAb(push, ctx->Depth.Test);
BEGIN_NV04(push, NV10_3D(DEPTH_WRITE_ENABLE), 1);
PUSH_DATAb(push, ctx->Depth.Mask);
BEGIN_NV04(push, NV10_3D(DEPTH_FUNC), 1);
PUSH_DATA (push, nvgl_comparison_op(ctx->Depth.Func));
}
static void
setup_combiners(NVPtr pNv, PicturePtr src, PicturePtr mask)
{
struct nouveau_channel *chan = pNv->chan;
uint32_t rc_in_alpha = 0, rc_in_rgb = 0;
if (PICT_FORMAT_A(src->format))
rc_in_alpha |= RC_IN_TEX(A, ALPHA, 0);
else
rc_in_alpha |= RC_IN_ONE(A);
if (mask && PICT_FORMAT_A(mask->format))
rc_in_alpha |= RC_IN_TEX(B, ALPHA, 1);
else
rc_in_alpha |= RC_IN_ONE(B);
if (effective_component_alpha(mask)) {
if (!needs_src_alpha(pNv->alu)) {
/* The alpha channels won't be used for blending. Drop
* them, as our pixels only have 4 components...
* output_i = src_i * mask_i
*/
if (PICT_FORMAT_RGB(src->format))
rc_in_rgb |= RC_IN_TEX(A, RGB, 0);
} else {
/* The RGB channels won't be used for blending. Drop
* them.
* output_i = src_alpha * mask_i
*/
if (PICT_FORMAT_A(src->format))
rc_in_rgb |= RC_IN_TEX(A, ALPHA, 0);
else
rc_in_rgb |= RC_IN_ONE(A);
}
rc_in_rgb |= RC_IN_TEX(B, RGB, 1);
} else {
if (PICT_FORMAT_RGB(src->format))
rc_in_rgb |= RC_IN_TEX(A, RGB, 0);
if (mask && PICT_FORMAT_A(mask->format))
rc_in_rgb |= RC_IN_TEX(B, ALPHA, 1);
else
rc_in_rgb |= RC_IN_ONE(B);
}
BEGIN_NV04(chan, NV10_3D(RC_IN_ALPHA(0)), 1);
OUT_RING (chan, rc_in_alpha);
BEGIN_NV04(chan, NV10_3D(RC_IN_RGB(0)), 1);
OUT_RING (chan, rc_in_rgb);
}
Bool
NV10EXAPrepareComposite(int op,
PicturePtr pict_src,
PicturePtr pict_mask,
PicturePtr pict_dst,
PixmapPtr src,
PixmapPtr mask,
PixmapPtr dst)
{
ScrnInfoPtr pScrn = xf86ScreenToScrn(dst->drawable.pScreen);
NVPtr pNv = NVPTR(pScrn);
struct nouveau_pushbuf *push = pNv->pushbuf;
uint32_t sc, sa, mc, ma;
if (!PUSH_SPACE(push, 128))
return FALSE;
PUSH_RESET(push);
/* setup render target and blending */
if (!setup_render_target(pNv, pict_dst, dst))
return FALSE;
setup_blend_function(pNv, pict_dst, op);
/* select picture sources */
if (!setup_picture(pNv, pict_src, src, 0, &sc, &sa))
return FALSE;
if (!setup_picture(pNv, pict_mask, mask, 1, &mc, &ma))
return FALSE;
/* configure register combiners */
BEGIN_NV04(push, NV10_3D(RC_IN_ALPHA(0)), 1);
PUSH_DATA (push, sa | ma);
BEGIN_NV04(push, NV10_3D(RC_IN_RGB(0)), 1);
if (effective_component_alpha(pict_mask)) {
if (needs_src_alpha(op))
PUSH_DATA(push, sa | mc);
else
PUSH_DATA(push, sc | mc);
} else {
PUSH_DATA(push, sc | ma);
}
nouveau_pushbuf_bufctx(push, pNv->bufctx);
if (nouveau_pushbuf_validate(push)) {
nouveau_pushbuf_bufctx(push, NULL);
return FALSE;
}
pNv->pspict = pict_src;
pNv->pmpict = pict_mask;
return TRUE;
}
void
nv10_emit_stencil_func(struct gl_context *ctx, int emit)
{
struct nouveau_pushbuf *push = context_push(ctx);
BEGIN_NV04(push, NV10_3D(STENCIL_ENABLE), 1);
PUSH_DATAb(push, ctx->Stencil.Enabled);
BEGIN_NV04(push, NV10_3D(STENCIL_FUNC_FUNC), 3);
PUSH_DATA (push, nvgl_comparison_op(ctx->Stencil.Function[0]));
PUSH_DATA (push, ctx->Stencil.Ref[0]);
PUSH_DATA (push, ctx->Stencil.ValueMask[0]);
}
static inline void
PUSH_VTX2s(struct nouveau_pushbuf *push,
int x1, int y1, int x2, int y2, int dx, int dy)
{
BEGIN_NV04(push, NV10_3D(VERTEX_TX0_2I), 1);
PUSH_DATA (push, ((y1 & 0xffff) << 16) | (x1 & 0xffff));
BEGIN_NV04(push, NV10_3D(VERTEX_TX1_2I), 1);
PUSH_DATA (push, ((y2 & 0xffff) << 16) | (x2 & 0xffff));
BEGIN_NV04(push, NV10_3D(VERTEX_POS_3F_X), 3);
PUSH_DATAf(push, dx);
PUSH_DATAf(push, dy);
PUSH_DATAf(push, 0.0);
}
static Bool
setup_render_target(NVPtr pNv, PicturePtr pict, PixmapPtr pixmap)
{
struct nouveau_pushbuf *push = pNv->pushbuf;
struct nouveau_bo *bo = nouveau_pixmap_bo(pixmap);
BEGIN_NV04(push, NV10_3D(RT_FORMAT), 3);
PUSH_DATA (push, get_rt_format(pict));
PUSH_DATA (push, (exaGetPixmapPitch(pixmap) << 16 |
exaGetPixmapPitch(pixmap)));
PUSH_MTHDl(push, NV10_3D(COLOR_OFFSET), bo, 0,
NOUVEAU_BO_VRAM | NOUVEAU_BO_RDWR);
return TRUE;
}
static void
nv10_render_bind_vertices(struct gl_context *ctx)
{
struct nouveau_render_state *render = to_render_state(ctx);
struct nouveau_pushbuf *push = context_push(ctx);
int i, attr;
FOR_EACH_BOUND_ATTR(render, i, attr) {
struct nouveau_array *a = &render->attrs[attr];
BEGIN_NV04(push, NV10_3D(VTXBUF_OFFSET(i)), 1);
PUSH_MTHDl(push, NV10_3D(VTXBUF_OFFSET(i)), BUFCTX_VTX,
a->bo, a->offset, NOUVEAU_BO_GART |
NOUVEAU_BO_RD);
}
}
static void
nv10_render_set_format(struct gl_context *ctx)
{
struct nouveau_render_state *render = to_render_state(ctx);
struct nouveau_pushbuf *push = context_push(ctx);
int i, attr, hw_format;
FOR_EACH_ATTR(render, i, attr) {
if (attr >= 0) {
struct nouveau_array *a = &render->attrs[attr];
hw_format = a->stride << 8 |
a->fields << 4 |
get_hw_format(a->type);
if (attr == VERT_ATTRIB_POS && a->fields == 4)
hw_format |= NV10_3D_VTXBUF_FMT_HOMOGENEOUS;
} else {
/* Unused attribute. */
hw_format = NV10_3D_VTXBUF_FMT_TYPE_V32_FLOAT;
}
BEGIN_NV04(push, NV10_3D(VTXBUF_FMT(i)), 1);
PUSH_DATA (push, hw_format);
}
}
void
nv10_emit_dither(struct gl_context *ctx, int emit)
{
struct nouveau_pushbuf *push = context_push(ctx);
BEGIN_NV04(push, NV10_3D(DITHER_ENABLE), 1);
PUSH_DATAb(push, ctx->Color.DitherFlag);
}
static Bool
setup_render_target(NVPtr pNv, PicturePtr pict, PixmapPtr pixmap)
{
struct nouveau_channel *chan = pNv->chan;
struct nouveau_bo *bo = nouveau_pixmap_bo(pixmap);
BEGIN_NV04(chan, NV10_3D(RT_FORMAT), 2);
OUT_RING (chan, get_rt_format(pict));
OUT_RING (chan, (exaGetPixmapPitch(pixmap) << 16 |
exaGetPixmapPitch(pixmap)));
BEGIN_NV04(chan, NV10_3D(COLOR_OFFSET), 1);
if (OUT_RELOCl(chan, bo, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR))
return FALSE;
return TRUE;
}
void
nv10_emit_stencil_mask(struct gl_context *ctx, int emit)
{
struct nouveau_pushbuf *push = context_push(ctx);
BEGIN_NV04(push, NV10_3D(STENCIL_MASK), 1);
PUSH_DATA (push, ctx->Stencil.WriteMask[0]);
}
void
nv10_emit_blend_func(struct gl_context *ctx, int emit)
{
struct nouveau_pushbuf *push = context_push(ctx);
BEGIN_NV04(push, NV10_3D(BLEND_FUNC_SRC), 2);
PUSH_DATA (push, nvgl_blend_func(ctx->Color.Blend[0].SrcRGB));
PUSH_DATA (push, nvgl_blend_func(ctx->Color.Blend[0].DstRGB));
}
void
nv10_emit_shade_model(struct gl_context *ctx, int emit)
{
struct nouveau_pushbuf *push = context_push(ctx);
BEGIN_NV04(push, NV10_3D(SHADE_MODEL), 1);
PUSH_DATA (push, ctx->Light.ShadeModel == GL_SMOOTH ?
NV10_3D_SHADE_MODEL_SMOOTH : NV10_3D_SHADE_MODEL_FLAT);
}
void
nv10_emit_stencil_op(struct gl_context *ctx, int emit)
{
struct nouveau_pushbuf *push = context_push(ctx);
BEGIN_NV04(push, NV10_3D(STENCIL_OP_FAIL), 3);
PUSH_DATA (push, nvgl_stencil_op(ctx->Stencil.FailFunc[0]));
PUSH_DATA (push, nvgl_stencil_op(ctx->Stencil.ZFailFunc[0]));
PUSH_DATA (push, nvgl_stencil_op(ctx->Stencil.ZPassFunc[0]));
}
void
nv10_emit_color_mask(struct gl_context *ctx, int emit)
{
struct nouveau_pushbuf *push = context_push(ctx);
BEGIN_NV04(push, NV10_3D(COLOR_MASK), 1);
PUSH_DATA (push, ((ctx->Color.ColorMask[0][3] ? 1 << 24 : 0) |
(ctx->Color.ColorMask[0][0] ? 1 << 16 : 0) |
(ctx->Color.ColorMask[0][1] ? 1 << 8 : 0) |
(ctx->Color.ColorMask[0][2] ? 1 << 0 : 0)));
}
void
nv10_emit_blend_color(struct gl_context *ctx, int emit)
{
struct nouveau_pushbuf *push = context_push(ctx);
BEGIN_NV04(push, NV10_3D(BLEND_COLOR), 1);
PUSH_DATA (push, FLOAT_TO_UBYTE(ctx->Color.BlendColor[3]) << 24 |
FLOAT_TO_UBYTE(ctx->Color.BlendColor[0]) << 16 |
FLOAT_TO_UBYTE(ctx->Color.BlendColor[1]) << 8 |
FLOAT_TO_UBYTE(ctx->Color.BlendColor[2]) << 0);
}
void
NV10EXAComposite(PixmapPtr pix_dst,
int sx, int sy, int mx, int my, int dx, int dy, int w, int h)
{
ScrnInfoPtr pScrn = xf86ScreenToScrn(pix_dst->drawable.pScreen);
NVPtr pNv = NVPTR(pScrn);
struct nouveau_pushbuf *push = pNv->pushbuf;
if (!PUSH_SPACE(push, 64))
return;
BEGIN_NV04(push, NV10_3D(VERTEX_BEGIN_END), 1);
PUSH_DATA (push, NV10_3D_VERTEX_BEGIN_END_QUADS);
PUSH_VTX2s(push, sx, sy, mx, my, dx, dy);
PUSH_VTX2s(push, sx + w, sy, mx + w, my, dx + w, dy);
PUSH_VTX2s(push, sx + w, sy + h, mx + w, my + h, dx + w, dy + h);
PUSH_VTX2s(push, sx, sy + h, mx, my + h, dx, dy + h);
BEGIN_NV04(push, NV10_3D(VERTEX_BEGIN_END), 1);
PUSH_DATA (push, NV10_3D_VERTEX_BEGIN_END_STOP);
}
static inline void
emit_vertex(NVPtr pNv, int i, PictVector pos[],
PictVector tex0[], PictVector tex1[])
{
struct nouveau_channel *chan = pNv->chan;
BEGIN_NV04(chan, NV10_3D(VERTEX_TX0_2F_S), 2);
OUT_RINGi (chan, tex0[i], 0);
OUT_RINGi (chan, tex0[i], 1);
if (tex1) {
BEGIN_NV04(chan, NV10_3D(VERTEX_TX1_2F_S), 2);
OUT_RINGi (chan, tex1[i], 0);
OUT_RINGi (chan, tex1[i], 1);
}
BEGIN_NV04(chan, NV10_3D(VERTEX_POS_3F_X), 3);
OUT_RINGi (chan, pos[i], 0);
OUT_RINGi (chan, pos[i], 1);
OUT_RINGf (chan, 0);
}
void
nv10_emit_tex_mat(struct gl_context *ctx, int emit)
{
const int i = emit - NOUVEAU_STATE_TEX_MAT0;
struct nouveau_context *nctx = to_nouveau_context(ctx);
struct nouveau_pushbuf *push = context_push(ctx);
if (nctx->fallback == HWTNL &&
((ctx->Texture._TexMatEnabled & 1 << i) ||
ctx->Texture.Unit[i]._GenFlags)) {
BEGIN_NV04(push, NV10_3D(TEX_MATRIX_ENABLE(i)), 1);
PUSH_DATA (push, 1);
BEGIN_NV04(push, NV10_3D(TEX_MATRIX(i, 0)), 16);
PUSH_DATAm(push, ctx->TextureMatrixStack[i].Top->m);
} else {
BEGIN_NV04(push, NV10_3D(TEX_MATRIX_ENABLE(i)), 1);
PUSH_DATA (push, 0);
}
}
static Bool
setup_texture(NVPtr pNv, int unit, PicturePtr pict, PixmapPtr pixmap)
{
struct nouveau_channel *chan = pNv->chan;
struct nouveau_bo *bo = nouveau_pixmap_bo(pixmap);
unsigned tex_reloc = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD;
long w = pict->pDrawable->width,
h = pict->pDrawable->height;
unsigned int txfmt =
NV10_3D_TEX_FORMAT_WRAP_T_CLAMP_TO_EDGE |
NV10_3D_TEX_FORMAT_WRAP_S_CLAMP_TO_EDGE |
log2i(w) << 20 | log2i(h) << 16 |
1 << 12 | /* lod == 1 */
get_tex_format(pict) |
0x50 /* UNK */;
BEGIN_NV04(chan, NV10_3D(TEX_OFFSET(unit)), 1);
if (OUT_RELOCl(chan, bo, 0, tex_reloc))
return FALSE;
if (pict->repeat == RepeatNone) {
/* NPOT_SIZE expects an even number for width, we can
* round up uneven numbers here because EXA always
* gives 64 byte aligned pixmaps and for all formats
* we support 64 bytes represents an even number of
* pixels
*/
w = (w + 1) &~ 1;
BEGIN_NV04(chan, NV10_3D(TEX_NPOT_PITCH(unit)), 1);
OUT_RING (chan, exaGetPixmapPitch(pixmap) << 16);
BEGIN_NV04(chan, NV10_3D(TEX_NPOT_SIZE(unit)), 1);
OUT_RING (chan, w << 16 | h);
}
BEGIN_NV04(chan, NV10_3D(TEX_FORMAT(unit)), 1 );
if (OUT_RELOCd(chan, bo, txfmt, tex_reloc | NOUVEAU_BO_OR,
NV10_3D_TEX_FORMAT_DMA0, NV10_3D_TEX_FORMAT_DMA1))
return FALSE;
BEGIN_NV04(chan, NV10_3D(TEX_ENABLE(unit)), 1 );
OUT_RING (chan, NV10_3D_TEX_ENABLE_ENABLE);
BEGIN_NV04(chan, NV10_3D(TEX_FILTER(unit)), 1);
if (pict->filter == PictFilterNearest)
OUT_RING(chan, (NV10_3D_TEX_FILTER_MAGNIFY_NEAREST |
NV10_3D_TEX_FILTER_MINIFY_NEAREST));
else
OUT_RING(chan, (NV10_3D_TEX_FILTER_MAGNIFY_LINEAR |
NV10_3D_TEX_FILTER_MINIFY_LINEAR));
return TRUE;
}
static Bool
setup_picture(NVPtr pNv, PicturePtr pict, PixmapPtr pixmap, int unit,
uint32_t *color, uint32_t *alpha)
{
struct nouveau_pushbuf *push = pNv->pushbuf;
uint32_t shift, source;
if (pict && pict->pDrawable) {
if (!setup_texture(pNv, unit, pict, pixmap))
return FALSE;
source = RCSRC_TEX(unit);
} else
if (pict) {
BEGIN_NV04(push, NV10_3D(RC_COLOR(unit)), 1);
PUSH_DATA (push, pict->pSourcePict->solidFill.color);
source = RCSRC_COL(unit);
}
if (pict && PICT_FORMAT_RGB(pict->format))
*color = RCSEL_COLOR | source;
else
*color = RCSEL_ALPHA | RCINP_ZERO;
if (pict && PICT_FORMAT_A(pict->format))
*alpha = RCSEL_ALPHA | source;
else
*alpha = RCSEL_ALPHA | RCINP_ONE;
if (unit)
shift = RCINP_B__SHIFT;
else
shift = RCINP_A__SHIFT;
*color <<= shift;
*alpha <<= shift;
return TRUE;
}
void
nv10_emit_tex_obj(struct gl_context *ctx, int emit)
{
const int i = emit - NOUVEAU_STATE_TEX_OBJ0;
struct nouveau_pushbuf *push = context_push(ctx);
const int bo_flags = NOUVEAU_BO_RD | NOUVEAU_BO_GART | NOUVEAU_BO_VRAM;
struct gl_texture_object *t;
struct nouveau_surface *s;
struct gl_texture_image *ti;
const struct gl_sampler_object *sa;
uint32_t tx_format, tx_filter, tx_enable;
PUSH_RESET(push, BUFCTX_TEX(i));
if (!ctx->Texture.Unit[i]._ReallyEnabled) {
BEGIN_NV04(push, NV10_3D(TEX_ENABLE(i)), 1);
PUSH_DATA (push, 0);
return;
}
t = ctx->Texture.Unit[i]._Current;
s = &to_nouveau_texture(t)->surfaces[t->BaseLevel];
ti = t->Image[0][t->BaseLevel];
sa = _mesa_get_samplerobj(ctx, i);
if (!nouveau_texture_validate(ctx, t))
return;
/* Recompute the texturing registers. */
tx_format = nvgl_wrap_mode(sa->WrapT) << 28
| nvgl_wrap_mode(sa->WrapS) << 24
| ti->HeightLog2 << 20
| ti->WidthLog2 << 16
| 5 << 4 | 1 << 12;
tx_filter = nvgl_filter_mode(sa->MagFilter) << 28
| nvgl_filter_mode(sa->MinFilter) << 24;
tx_enable = NV10_3D_TEX_ENABLE_ENABLE
| log2i(sa->MaxAnisotropy) << 4;
if (t->Target == GL_TEXTURE_RECTANGLE) {
BEGIN_NV04(push, NV10_3D(TEX_NPOT_PITCH(i)), 1);
PUSH_DATA (push, s->pitch << 16);
BEGIN_NV04(push, NV10_3D(TEX_NPOT_SIZE(i)), 1);
PUSH_DATA (push, align(s->width, 2) << 16 | s->height);
tx_format |= get_tex_format_rect(ti);
} else {
tx_format |= get_tex_format_pot(ti);
}
if (sa->MinFilter != GL_NEAREST &&
sa->MinFilter != GL_LINEAR) {
int lod_min = sa->MinLod;
int lod_max = MIN2(sa->MaxLod, t->_MaxLambda);
int lod_bias = sa->LodBias
+ ctx->Texture.Unit[i].LodBias;
lod_max = CLAMP(lod_max, 0, 15);
lod_min = CLAMP(lod_min, 0, 15);
lod_bias = CLAMP(lod_bias, 0, 15);
tx_format |= NV10_3D_TEX_FORMAT_MIPMAP;
tx_filter |= lod_bias << 8;
tx_enable |= lod_min << 26
| lod_max << 14;
}
/* Write it to the hardware. */
BEGIN_NV04(push, NV10_3D(TEX_FORMAT(i)), 1);
PUSH_MTHD (push, NV10_3D(TEX_FORMAT(i)), BUFCTX_TEX(i),
s->bo, tx_format, bo_flags | NOUVEAU_BO_OR,
NV10_3D_TEX_FORMAT_DMA0,
NV10_3D_TEX_FORMAT_DMA1);
BEGIN_NV04(push, NV10_3D(TEX_OFFSET(i)), 1);
PUSH_MTHDl(push, NV10_3D(TEX_OFFSET(i)), BUFCTX_TEX(i),
s->bo, s->offset, bo_flags);
BEGIN_NV04(push, NV10_3D(TEX_FILTER(i)), 1);
PUSH_DATA (push, tx_filter);
BEGIN_NV04(push, NV10_3D(TEX_ENABLE(i)), 1);
PUSH_DATA (push, tx_enable);
}
static Bool
setup_texture(NVPtr pNv, int unit, PicturePtr pict, PixmapPtr pixmap)
{
struct nouveau_pushbuf *push = pNv->pushbuf;
struct nouveau_bo *bo = nouveau_pixmap_bo(pixmap);
unsigned reloc = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD;
unsigned h = pict->pDrawable->height;
unsigned w = pict->pDrawable->width;
unsigned format;
format = NV10_3D_TEX_FORMAT_WRAP_T_CLAMP_TO_EDGE |
NV10_3D_TEX_FORMAT_WRAP_S_CLAMP_TO_EDGE |
log2i(w) << 20 | log2i(h) << 16 |
1 << 12 | /* lod == 1 */
get_tex_format(pNv, pict) |
0x50 /* UNK */;
/* NPOT_SIZE expects an even number for width, we can round up uneven
* numbers here because EXA always gives 64 byte aligned pixmaps and
* for all formats we support 64 bytes represents an even number of
* pixels
*/
w = (w + 1) & ~1;
BEGIN_NV04(push, NV10_3D(TEX_OFFSET(unit)), 1);
PUSH_MTHDl(push, NV10_3D(TEX_OFFSET(unit)), bo, 0, reloc);
BEGIN_NV04(push, NV10_3D(TEX_FORMAT(unit)), 1);
PUSH_MTHDs(push, NV10_3D(TEX_FORMAT(unit)), bo, format, reloc,
NV10_3D_TEX_FORMAT_DMA0,
NV10_3D_TEX_FORMAT_DMA1);
BEGIN_NV04(push, NV10_3D(TEX_ENABLE(unit)), 1 );
PUSH_DATA (push, NV10_3D_TEX_ENABLE_ENABLE);
BEGIN_NV04(push, NV10_3D(TEX_NPOT_PITCH(unit)), 1);
PUSH_DATA (push, exaGetPixmapPitch(pixmap) << 16);
BEGIN_NV04(push, NV10_3D(TEX_NPOT_SIZE(unit)), 1);
PUSH_DATA (push, (w << 16) | h);
BEGIN_NV04(push, NV10_3D(TEX_FILTER(unit)), 1);
if (pict->filter == PictFilterNearest)
PUSH_DATA(push, NV10_3D_TEX_FILTER_MAGNIFY_NEAREST |
NV10_3D_TEX_FILTER_MINIFY_NEAREST);
else
PUSH_DATA(push, NV10_3D_TEX_FILTER_MAGNIFY_LINEAR |
NV10_3D_TEX_FILTER_MINIFY_LINEAR);
if (pict->transform) {
BEGIN_NV04(push, NV10_3D(TEX_MATRIX_ENABLE(unit)), 1);
PUSH_DATA (push, 1);
BEGIN_NV04(push, NV10_3D(TEX_MATRIX(unit, 0)), 16);
PUSH_DATAf(push, xFixedToFloat(pict->transform->matrix[0][0]));
PUSH_DATAf(push, xFixedToFloat(pict->transform->matrix[0][1]));
PUSH_DATAf(push, 0.f);
PUSH_DATAf(push, xFixedToFloat(pict->transform->matrix[0][2]));
PUSH_DATAf(push, xFixedToFloat(pict->transform->matrix[1][0]));
PUSH_DATAf(push, xFixedToFloat(pict->transform->matrix[1][1]));
PUSH_DATAf(push, 0.f);
PUSH_DATAf(push, xFixedToFloat(pict->transform->matrix[1][2]));
PUSH_DATAf(push, 0.0f);
PUSH_DATAf(push, 0.0f);
PUSH_DATAf(push, 0.0f);
PUSH_DATAf(push, 0.0f);
PUSH_DATAf(push, xFixedToFloat(pict->transform->matrix[2][0]));
PUSH_DATAf(push, xFixedToFloat(pict->transform->matrix[2][1]));
PUSH_DATAf(push, 0.0f);
PUSH_DATAf(push, xFixedToFloat(pict->transform->matrix[2][2]));
} else {
BEGIN_NV04(push, NV10_3D(TEX_MATRIX_ENABLE(unit)), 1);
PUSH_DATA (push, 0);
}
return TRUE;
}
static void
nv10_hwctx_init(struct gl_context *ctx)
{
struct nouveau_pushbuf *push = context_push(ctx);
struct nouveau_hw_state *hw = &to_nouveau_context(ctx)->hw;
struct nv04_fifo *fifo = hw->chan->data;
int i;
BEGIN_NV04(push, NV01_SUBC(3D, OBJECT), 1);
PUSH_DATA (push, hw->eng3d->handle);
BEGIN_NV04(push, NV10_3D(DMA_NOTIFY), 1);
PUSH_DATA (push, hw->ntfy->handle);
BEGIN_NV04(push, NV10_3D(DMA_TEXTURE0), 3);
PUSH_DATA (push, fifo->vram);
PUSH_DATA (push, fifo->gart);
PUSH_DATA (push, fifo->gart);
BEGIN_NV04(push, NV10_3D(DMA_COLOR), 2);
PUSH_DATA (push, fifo->vram);
PUSH_DATA (push, fifo->vram);
BEGIN_NV04(push, NV04_GRAPH(3D, NOP), 1);
PUSH_DATA (push, 0);
BEGIN_NV04(push, NV10_3D(RT_HORIZ), 2);
PUSH_DATA (push, 0);
PUSH_DATA (push, 0);
BEGIN_NV04(push, NV10_3D(VIEWPORT_CLIP_HORIZ(0)), 1);
PUSH_DATA (push, 0x7ff << 16 | 0x800);
BEGIN_NV04(push, NV10_3D(VIEWPORT_CLIP_VERT(0)), 1);
PUSH_DATA (push, 0x7ff << 16 | 0x800);
for (i = 1; i < 8; i++) {
BEGIN_NV04(push, NV10_3D(VIEWPORT_CLIP_HORIZ(i)), 1);
PUSH_DATA (push, 0);
BEGIN_NV04(push, NV10_3D(VIEWPORT_CLIP_VERT(i)), 1);
PUSH_DATA (push, 0);
}
BEGIN_NV04(push, SUBC_3D(0x290), 1);
PUSH_DATA (push, 0x10 << 16 | 1);
BEGIN_NV04(push, SUBC_3D(0x3f4), 1);
PUSH_DATA (push, 0);
BEGIN_NV04(push, NV04_GRAPH(3D, NOP), 1);
PUSH_DATA (push, 0);
if (context_chipset(ctx) >= 0x17) {
BEGIN_NV04(push, NV17_3D(UNK01AC), 2);
PUSH_DATA (push, fifo->vram);
PUSH_DATA (push, fifo->vram);
BEGIN_NV04(push, SUBC_3D(0xd84), 1);
PUSH_DATA (push, 0x3);
BEGIN_NV04(push, NV17_3D(COLOR_MASK_ENABLE), 1);
PUSH_DATA (push, 1);
}
if (context_chipset(ctx) >= 0x11) {
BEGIN_NV04(push, SUBC_3D(0x120), 3);
PUSH_DATA (push, 0);
PUSH_DATA (push, 1);
PUSH_DATA (push, 2);
BEGIN_NV04(push, NV04_GRAPH(3D, NOP), 1);
PUSH_DATA (push, 0);
}
BEGIN_NV04(push, NV04_GRAPH(3D, NOP), 1);
PUSH_DATA (push, 0);
/* Set state */
BEGIN_NV04(push, NV10_3D(FOG_ENABLE), 1);
PUSH_DATA (push, 0);
BEGIN_NV04(push, NV10_3D(ALPHA_FUNC_ENABLE), 1);
PUSH_DATA (push, 0);
BEGIN_NV04(push, NV10_3D(ALPHA_FUNC_FUNC), 2);
PUSH_DATA (push, 0x207);
PUSH_DATA (push, 0);
BEGIN_NV04(push, NV10_3D(TEX_ENABLE(0)), 2);
PUSH_DATA (push, 0);
PUSH_DATA (push, 0);
BEGIN_NV04(push, NV10_3D(BLEND_FUNC_ENABLE), 1);
PUSH_DATA (push, 0);
BEGIN_NV04(push, NV10_3D(DITHER_ENABLE), 2);
PUSH_DATA (push, 1);
PUSH_DATA (push, 0);
BEGIN_NV04(push, NV10_3D(LINE_SMOOTH_ENABLE), 1);
PUSH_DATA (push, 0);
BEGIN_NV04(push, NV10_3D(VERTEX_WEIGHT_ENABLE), 2);
PUSH_DATA (push, 0);
PUSH_DATA (push, 0);
BEGIN_NV04(push, NV10_3D(BLEND_FUNC_SRC), 4);
PUSH_DATA (push, 1);
PUSH_DATA (push, 0);
PUSH_DATA (push, 0);
PUSH_DATA (push, 0x8006);
BEGIN_NV04(push, NV10_3D(STENCIL_MASK), 8);
PUSH_DATA (push, 0xff);
PUSH_DATA (push, 0x207);
PUSH_DATA (push, 0);
PUSH_DATA (push, 0xff);
PUSH_DATA (push, 0x1e00);
PUSH_DATA (push, 0x1e00);
PUSH_DATA (push, 0x1e00);
PUSH_DATA (push, 0x1d01);
BEGIN_NV04(push, NV10_3D(NORMALIZE_ENABLE), 1);
PUSH_DATA (push, 0);
BEGIN_NV04(push, NV10_3D(FOG_ENABLE), 2);
PUSH_DATA (push, 0);
PUSH_DATA (push, 0);
BEGIN_NV04(push, NV10_3D(LIGHT_MODEL), 1);
PUSH_DATA (push, 0);
BEGIN_NV04(push, NV10_3D(SEPARATE_SPECULAR_ENABLE), 1);
PUSH_DATA (push, 0);
BEGIN_NV04(push, NV10_3D(ENABLED_LIGHTS), 1);
PUSH_DATA (push, 0);
BEGIN_NV04(push, NV10_3D(POLYGON_OFFSET_POINT_ENABLE), 3);
PUSH_DATA (push, 0);
PUSH_DATA (push, 0);
PUSH_DATA (push, 0);
BEGIN_NV04(push, NV10_3D(DEPTH_FUNC), 1);
PUSH_DATA (push, 0x201);
BEGIN_NV04(push, NV10_3D(DEPTH_WRITE_ENABLE), 1);
PUSH_DATA (push, 0);
BEGIN_NV04(push, NV10_3D(DEPTH_TEST_ENABLE), 1);
PUSH_DATA (push, 0);
BEGIN_NV04(push, NV10_3D(POLYGON_OFFSET_FACTOR), 2);
PUSH_DATA (push, 0);
PUSH_DATA (push, 0);
BEGIN_NV04(push, NV10_3D(POINT_SIZE), 1);
PUSH_DATA (push, 8);
BEGIN_NV04(push, NV10_3D(POINT_PARAMETERS_ENABLE), 2);
PUSH_DATA (push, 0);
PUSH_DATA (push, 0);
BEGIN_NV04(push, NV10_3D(LINE_WIDTH), 1);
PUSH_DATA (push, 8);
BEGIN_NV04(push, NV10_3D(LINE_SMOOTH_ENABLE), 1);
PUSH_DATA (push, 0);
BEGIN_NV04(push, NV10_3D(POLYGON_MODE_FRONT), 2);
PUSH_DATA (push, 0x1b02);
PUSH_DATA (push, 0x1b02);
BEGIN_NV04(push, NV10_3D(CULL_FACE), 2);
PUSH_DATA (push, 0x405);
PUSH_DATA (push, 0x901);
BEGIN_NV04(push, NV10_3D(POLYGON_SMOOTH_ENABLE), 1);
PUSH_DATA (push, 0);
BEGIN_NV04(push, NV10_3D(CULL_FACE_ENABLE), 1);
PUSH_DATA (push, 0);
BEGIN_NV04(push, NV10_3D(TEX_GEN_MODE(0, 0)), 8);
for (i = 0; i < 8; i++)
PUSH_DATA (push, 0);
BEGIN_NV04(push, NV10_3D(TEX_MATRIX_ENABLE(0)), 2);
PUSH_DATA (push, 0);
PUSH_DATA (push, 0);
BEGIN_NV04(push, NV10_3D(FOG_COEFF(0)), 3);
PUSH_DATA (push, 0x3fc00000); /* -1.50 */
PUSH_DATA (push, 0xbdb8aa0a); /* -0.09 */
PUSH_DATA (push, 0); /* 0.00 */
BEGIN_NV04(push, NV04_GRAPH(3D, NOP), 1);
PUSH_DATA (push, 0);
BEGIN_NV04(push, NV10_3D(FOG_MODE), 2);
PUSH_DATA (push, 0x802);
PUSH_DATA (push, 2);
/* for some reason VIEW_MATRIX_ENABLE need to be 6 instead of 4 when
* using texturing, except when using the texture matrix
*/
BEGIN_NV04(push, NV10_3D(VIEW_MATRIX_ENABLE), 1);
PUSH_DATA (push, 6);
BEGIN_NV04(push, NV10_3D(COLOR_MASK), 1);
PUSH_DATA (push, 0x01010101);
/* Set vertex component */
BEGIN_NV04(push, NV10_3D(VERTEX_COL_4F_R), 4);
PUSH_DATAf(push, 1.0);
PUSH_DATAf(push, 0.0);
PUSH_DATAf(push, 0.0);
PUSH_DATAf(push, 1.0);
BEGIN_NV04(push, NV10_3D(VERTEX_COL2_3F_R), 3);
PUSH_DATA (push, 0);
PUSH_DATA (push, 0);
PUSH_DATA (push, 0);
BEGIN_NV04(push, NV10_3D(VERTEX_NOR_3F_X), 3);
PUSH_DATA (push, 0);
PUSH_DATA (push, 0);
PUSH_DATAf(push, 1.0);
BEGIN_NV04(push, NV10_3D(VERTEX_TX0_4F_S), 4);
PUSH_DATAf(push, 0.0);
PUSH_DATAf(push, 0.0);
PUSH_DATAf(push, 0.0);
PUSH_DATAf(push, 1.0);
BEGIN_NV04(push, NV10_3D(VERTEX_TX1_4F_S), 4);
PUSH_DATAf(push, 0.0);
PUSH_DATAf(push, 0.0);
PUSH_DATAf(push, 0.0);
PUSH_DATAf(push, 1.0);
BEGIN_NV04(push, NV10_3D(VERTEX_FOG_1F), 1);
PUSH_DATAf(push, 0.0);
BEGIN_NV04(push, NV10_3D(EDGEFLAG_ENABLE), 1);
PUSH_DATA (push, 1);
BEGIN_NV04(push, NV10_3D(DEPTH_RANGE_NEAR), 2);
PUSH_DATAf(push, 0.0);
PUSH_DATAf(push, 16777216.0);
PUSH_KICK (push);
}
static void
setup_hierz_buffer(struct gl_context *ctx)
{
struct nouveau_pushbuf *push = context_push(ctx);
struct gl_framebuffer *fb = ctx->DrawBuffer;
struct nouveau_framebuffer *nfb = to_nouveau_framebuffer(fb);
unsigned pitch = align(fb->Width, 128),
height = align(fb->Height, 2),
size = pitch * height;
if (!nfb->hierz.bo || nfb->hierz.bo->size != size) {
union nouveau_bo_config config = {
.nv04.surf_flags = NV04_BO_ZETA,
.nv04.surf_pitch = 0
};
nouveau_bo_ref(NULL, &nfb->hierz.bo);
nouveau_bo_new(context_dev(ctx), NOUVEAU_BO_VRAM, 0, size,
&config, &nfb->hierz.bo);
}
PUSH_SPACE(push, 11);
BEGIN_NV04(push, NV17_3D(HIERZ_OFFSET), 1);
PUSH_MTHDl(push, NV17_3D(HIERZ_OFFSET), BUFCTX_FB,
nfb->hierz.bo, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_RDWR);
BEGIN_NV04(push, NV17_3D(HIERZ_WINDOW_X), 4);
PUSH_DATAf(push, - 1792);
PUSH_DATAf(push, - 2304 + fb->Height);
PUSH_DATAf(push, fb->_DepthMaxF / 2);
PUSH_DATAf(push, 0);
BEGIN_NV04(push, NV17_3D(HIERZ_PITCH), 1);
PUSH_DATA (push, pitch);
BEGIN_NV04(push, NV17_3D(HIERZ_ENABLE), 1);
PUSH_DATA (push, 1);
}
void
nv10_emit_framebuffer(struct gl_context *ctx, int emit)
{
struct nouveau_pushbuf *push = context_push(ctx);
struct gl_framebuffer *fb = ctx->DrawBuffer;
struct nouveau_surface *s;
unsigned rt_format = NV10_3D_RT_FORMAT_TYPE_LINEAR;
unsigned rt_pitch = 0, zeta_pitch = 0;
unsigned bo_flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_RDWR;
if (fb->_Status != GL_FRAMEBUFFER_COMPLETE_EXT)
return;
PUSH_RESET(push, BUFCTX_FB);
/* At least nv11 seems to get sad if we don't do this before
* swapping RTs.*/
if (context_eng3d(ctx)->oclass < NV17_3D_CLASS) {
int i;
for (i = 0; i < 6; i++) {
BEGIN_NV04(push, NV04_GRAPH(3D, NOP), 1);
PUSH_DATA (push, 0);
}
}
/* Render target */
if (fb->_ColorDrawBuffers[0]) {
s = &to_nouveau_renderbuffer(
fb->_ColorDrawBuffers[0])->surface;
rt_format |= get_rt_format(s->format);
zeta_pitch = rt_pitch = s->pitch;
BEGIN_NV04(push, NV10_3D(COLOR_OFFSET), 1);
PUSH_MTHDl(push, NV10_3D(COLOR_OFFSET), BUFCTX_FB,
s->bo, 0, bo_flags);
}
/* depth/stencil */
if (fb->Attachment[BUFFER_DEPTH].Renderbuffer) {
s = &to_nouveau_renderbuffer(
fb->Attachment[BUFFER_DEPTH].Renderbuffer)->surface;
rt_format |= get_rt_format(s->format);
zeta_pitch = s->pitch;
BEGIN_NV04(push, NV10_3D(ZETA_OFFSET), 1);
PUSH_MTHDl(push, NV10_3D(ZETA_OFFSET), BUFCTX_FB,
s->bo, 0, bo_flags);
if (context_eng3d(ctx)->oclass >= NV17_3D_CLASS) {
setup_hierz_buffer(ctx);
context_dirty(ctx, ZCLEAR);
}
}
BEGIN_NV04(push, NV10_3D(RT_FORMAT), 2);
PUSH_DATA (push, rt_format);
PUSH_DATA (push, zeta_pitch << 16 | rt_pitch);
context_dirty(ctx, VIEWPORT);
context_dirty(ctx, SCISSOR);
context_dirty(ctx, DEPTH);
}
void
nv10_emit_render_mode(struct gl_context *ctx, int emit)
{
}
void
nv10_emit_scissor(struct gl_context *ctx, int emit)
{
struct nouveau_pushbuf *push = context_push(ctx);
int x, y, w, h;
get_scissors(ctx->DrawBuffer, &x, &y, &w, &h);
BEGIN_NV04(push, NV10_3D(RT_HORIZ), 2);
PUSH_DATA (push, w << 16 | x);
PUSH_DATA (push, h << 16 | y);
}
void
nv10_emit_viewport(struct gl_context *ctx, int emit)
{
struct nouveau_pushbuf *push = context_push(ctx);
struct gl_viewport_attrib *vp = &ctx->ViewportArray[0];
struct gl_framebuffer *fb = ctx->DrawBuffer;
float a[4] = {};
get_viewport_translate(ctx, a);
a[0] -= 2048;
a[1] -= 2048;
if (nv10_use_viewport_zclear(ctx))
a[2] = nv10_transform_depth(ctx, (vp->Far + vp->Near) / 2);
BEGIN_NV04(push, NV10_3D(VIEWPORT_TRANSLATE_X), 4);
PUSH_DATAp(push, a, 4);
BEGIN_NV04(push, NV10_3D(VIEWPORT_CLIP_HORIZ(0)), 1);
PUSH_DATA (push, (fb->Width - 1) << 16 | 0x08000800);
BEGIN_NV04(push, NV10_3D(VIEWPORT_CLIP_VERT(0)), 1);
PUSH_DATA (push, (fb->Height - 1) << 16 | 0x08000800);
context_dirty(ctx, PROJECTION);
}
void
nv10_emit_zclear(struct gl_context *ctx, int emit)
{
struct nouveau_context *nctx = to_nouveau_context(ctx);
struct nouveau_pushbuf *push = context_push(ctx);
struct nouveau_framebuffer *nfb =
to_nouveau_framebuffer(ctx->DrawBuffer);
if (nfb->hierz.bo) {
BEGIN_NV04(push, NV17_3D(ZCLEAR_ENABLE), 2);
PUSH_DATAb(push, !nctx->hierz.clear_blocked);
PUSH_DATA (push, nfb->hierz.clear_value |
(nctx->hierz.clear_seq & 0xff));
} else {
BEGIN_NV04(push, NV10_3D(DEPTH_RANGE_NEAR), 2);
PUSH_DATAf(push, nv10_transform_depth(ctx, 0));
PUSH_DATAf(push, nv10_transform_depth(ctx, 1));
context_dirty(ctx, VIEWPORT);
}
}