#include "util/u_format.h"
#include "util/u_surface.h"
+#include "tgsi/tgsi_ureg.h"
+
+#include "os/os_thread.h"
+
#include "nv50_context.h"
#include "nv50_resource.h"
+#include "nv50_blit.h"
#include "nv50_defs.xml.h"
+#include "nv50_texture.xml.h"
#define NV50_ENG2D_SUPPORTED_FORMATS 0xff0843e080608409ULL
}
static int
-nv50_2d_texture_set(struct nouveau_channel *chan, int dst,
- struct nv50_miptree *mt, unsigned level, unsigned layer)
+nv50_2d_texture_set(struct nouveau_pushbuf *push, int dst,
+ struct nv50_miptree *mt, unsigned level, unsigned layer,
+ enum pipe_format pformat)
{
struct nouveau_bo *bo = mt->base.bo;
uint32_t width, height, depth;
uint32_t format;
uint32_t mthd = dst ? NV50_2D_DST_FORMAT : NV50_2D_SRC_FORMAT;
- uint32_t flags = mt->base.domain | (dst ? NOUVEAU_BO_WR : NOUVEAU_BO_RD);
uint32_t offset = mt->level[level].offset;
- format = nv50_2d_format(mt->base.base.format);
+ format = nv50_2d_format(pformat);
if (!format) {
NOUVEAU_ERR("invalid/unsupported surface format: %s\n",
- util_format_name(mt->base.base.format));
+ util_format_name(pformat));
return 1;
}
width = u_minify(mt->base.base.width0, level) << mt->ms_x;
height = u_minify(mt->base.base.height0, level) << mt->ms_y;
+ depth = u_minify(mt->base.base.depth0, level);
offset = mt->level[level].offset;
if (!mt->layout_3d) {
offset += mt->layer_stride * layer;
depth = 1;
layer = 0;
+ } else
+ if (!dst) {
+ offset += nv50_mt_zslice_offset(mt, level, layer);
+ layer = 0;
+ }
+
+ if (!nouveau_bo_memtype(bo)) {
+ BEGIN_NV04(push, SUBC_2D(mthd), 2);
+ PUSH_DATA (push, format);
+ PUSH_DATA (push, 1);
+ BEGIN_NV04(push, SUBC_2D(mthd + 0x14), 5);
+ PUSH_DATA (push, mt->level[level].pitch);
+ PUSH_DATA (push, width);
+ PUSH_DATA (push, height);
+ PUSH_DATAh(push, bo->offset + offset);
+ PUSH_DATA (push, bo->offset + offset);
} else {
- depth = u_minify(mt->base.base.depth0, level);
- }
-
- if (!(bo->tile_flags & NOUVEAU_BO_TILE_LAYOUT_MASK)) {
- BEGIN_RING(chan, RING_2D_(mthd), 2);
- OUT_RING (chan, format);
- OUT_RING (chan, 1);
- BEGIN_RING(chan, RING_2D_(mthd + 0x14), 5);
- OUT_RING (chan, mt->level[level].pitch);
- OUT_RING (chan, width);
- OUT_RING (chan, height);
- OUT_RELOCh(chan, bo, offset, flags);
- OUT_RELOCl(chan, bo, offset, flags);
- } else {
- BEGIN_RING(chan, RING_2D_(mthd), 5);
- OUT_RING (chan, format);
- OUT_RING (chan, 0);
- OUT_RING (chan, mt->level[level].tile_mode << 4);
- OUT_RING (chan, depth);
- OUT_RING (chan, layer);
- BEGIN_RING(chan, RING_2D_(mthd + 0x18), 4);
- OUT_RING (chan, width);
- OUT_RING (chan, height);
- OUT_RELOCh(chan, bo, offset, flags);
- OUT_RELOCl(chan, bo, offset, flags);
+ BEGIN_NV04(push, SUBC_2D(mthd), 5);
+ PUSH_DATA (push, format);
+ PUSH_DATA (push, 0);
+ PUSH_DATA (push, mt->level[level].tile_mode);
+ PUSH_DATA (push, depth);
+ PUSH_DATA (push, layer);
+ BEGIN_NV04(push, SUBC_2D(mthd + 0x18), 4);
+ PUSH_DATA (push, width);
+ PUSH_DATA (push, height);
+ PUSH_DATAh(push, bo->offset + offset);
+ PUSH_DATA (push, bo->offset + offset);
}
#if 0
if (dst) {
- BEGIN_RING(chan, RING_2D_(NV50_2D_CLIP_X), 4);
- OUT_RING (chan, 0);
- OUT_RING (chan, 0);
- OUT_RING (chan, width);
- OUT_RING (chan, height);
+ BEGIN_NV04(push, SUBC_2D(NV50_2D_CLIP_X), 4);
+ PUSH_DATA (push, 0);
+ PUSH_DATA (push, 0);
+ PUSH_DATA (push, width);
+ PUSH_DATA (push, height);
}
#endif
return 0;
}
static int
-nv50_2d_texture_do_copy(struct nouveau_channel *chan,
+nv50_2d_texture_do_copy(struct nouveau_pushbuf *push,
struct nv50_miptree *dst, unsigned dst_level,
unsigned dx, unsigned dy, unsigned dz,
struct nv50_miptree *src, unsigned src_level,
unsigned sx, unsigned sy, unsigned sz,
unsigned w, unsigned h)
{
- static const uint32_t duvdxy[5] =
- {
- 0x40000000, 0x80000000, 0x00000001, 0x00000002, 0x00000004
- };
-
+ const enum pipe_format dfmt = dst->base.base.format;
+ const enum pipe_format sfmt = src->base.base.format;
int ret;
- uint32_t ctrl;
- ret = MARK_RING(chan, 2 * 16 + 32, 4);
- if (ret)
- return ret;
+ if (!PUSH_SPACE(push, 2 * 16 + 32))
+ return PIPE_ERROR;
- ret = nv50_2d_texture_set(chan, 1, dst, dst_level, dz);
+ ret = nv50_2d_texture_set(push, 1, dst, dst_level, dz, dfmt);
if (ret)
return ret;
- ret = nv50_2d_texture_set(chan, 0, src, src_level, sz);
+ ret = nv50_2d_texture_set(push, 0, src, src_level, sz, sfmt);
if (ret)
return ret;
- /* NOTE: 2D engine doesn't work for MS8 */
- if (src->ms_x)
- ctrl = 0x11;
-
- /* 0/1 = CENTER/CORNER, 00/10 = POINT/BILINEAR */
- BEGIN_RING(chan, RING_2D(BLIT_CONTROL), 1);
- OUT_RING (chan, ctrl);
- BEGIN_RING(chan, RING_2D(BLIT_DST_X), 4);
- OUT_RING (chan, dx << dst->ms_x);
- OUT_RING (chan, dy << dst->ms_y);
- OUT_RING (chan, w << dst->ms_x);
- OUT_RING (chan, h << dst->ms_y);
- BEGIN_RING(chan, RING_2D(BLIT_DU_DX_FRACT), 4);
- OUT_RING (chan, duvdxy[2 + ((int)src->ms_x - (int)dst->ms_x)] & 0xf0000000);
- OUT_RING (chan, duvdxy[2 + ((int)src->ms_x - (int)dst->ms_x)] & 0x0000000f);
- OUT_RING (chan, duvdxy[2 + ((int)src->ms_y - (int)dst->ms_y)] & 0xf0000000);
- OUT_RING (chan, duvdxy[2 + ((int)src->ms_y - (int)dst->ms_y)] & 0x0000000f);
- BEGIN_RING(chan, RING_2D(BLIT_SRC_X_FRACT), 4);
- OUT_RING (chan, 0);
- OUT_RING (chan, sx << src->ms_x);
- OUT_RING (chan, 0);
- OUT_RING (chan, sy << src->ms_y);
+ BEGIN_NV04(push, NV50_2D(BLIT_CONTROL), 1);
+ PUSH_DATA (push, NV50_2D_BLIT_CONTROL_FILTER_POINT_SAMPLE);
+ BEGIN_NV04(push, NV50_2D(BLIT_DST_X), 4);
+ PUSH_DATA (push, dx << dst->ms_x);
+ PUSH_DATA (push, dy << dst->ms_y);
+ PUSH_DATA (push, w << dst->ms_x);
+ PUSH_DATA (push, h << dst->ms_y);
+ BEGIN_NV04(push, NV50_2D(BLIT_DU_DX_FRACT), 4);
+ PUSH_DATA (push, 0);
+ PUSH_DATA (push, 1);
+ PUSH_DATA (push, 0);
+ PUSH_DATA (push, 1);
+ BEGIN_NV04(push, NV50_2D(BLIT_SRC_X_FRACT), 4);
+ PUSH_DATA (push, 0);
+ PUSH_DATA (push, sx << src->ms_x);
+ PUSH_DATA (push, 0);
+ PUSH_DATA (push, sy << src->ms_y);
return 0;
}
struct pipe_resource *src, unsigned src_level,
const struct pipe_box *src_box)
{
- struct nv50_screen *screen = nv50_context(pipe)->screen;
+ struct nv50_context *nv50 = nv50_context(pipe);
int ret;
boolean m2mf;
unsigned dst_layer = dstz, src_layer = src_box->z;
return;
}
- assert(src->nr_samples == dst->nr_samples);
+ /* 0 and 1 are equal, only supporting 0/1, 2, 4 and 8 */
+ assert((src->nr_samples | 1) == (dst->nr_samples | 1));
m2mf = (src->format == dst->format) ||
(util_format_get_blocksizebits(src->format) ==
src_box->x, src_box->y, src_box->z);
for (i = 0; i < src_box->depth; ++i) {
- nv50_m2mf_transfer_rect(&screen->base.base, &drect, &srect, nx, ny);
+ nv50_m2mf_transfer_rect(nv50, &drect, &srect, nx, ny);
if (nv50_miptree(dst)->layout_3d)
drect.z++;
(nv50_2d_format_faithful(src->format) &&
nv50_2d_format_faithful(dst->format)));
+ BCTX_REFN(nv50->bufctx, 2D, nv04_resource(src), RD);
+ BCTX_REFN(nv50->bufctx, 2D, nv04_resource(dst), WR);
+ nouveau_pushbuf_bufctx(nv50->base.pushbuf, nv50->bufctx);
+ nouveau_pushbuf_validate(nv50->base.pushbuf);
+
for (; dst_layer < dstz + src_box->depth; ++dst_layer, ++src_layer) {
- ret = nv50_2d_texture_do_copy(screen->base.channel,
+ ret = nv50_2d_texture_do_copy(nv50->base.pushbuf,
nv50_miptree(dst), dst_level,
dstx, dsty, dst_layer,
nv50_miptree(src), src_level,
src_box->x, src_box->y, src_layer,
src_box->width, src_box->height);
if (ret)
- return;
+ break;
}
+ nouveau_bufctx_reset(nv50->bufctx, NV50_BIND_2D);
}
static void
unsigned width, unsigned height)
{
struct nv50_context *nv50 = nv50_context(pipe);
- struct nv50_screen *screen = nv50->screen;
- struct nouveau_channel *chan = screen->base.channel;
+ struct nouveau_pushbuf *push = nv50->base.pushbuf;
struct nv50_miptree *mt = nv50_miptree(dst->texture);
struct nv50_surface *sf = nv50_surface(dst);
struct nouveau_bo *bo = mt->base.bo;
+ unsigned z;
- BEGIN_RING(chan, RING_3D(CLEAR_COLOR(0)), 4);
- OUT_RINGf (chan, color->f[0]);
- OUT_RINGf (chan, color->f[1]);
- OUT_RINGf (chan, color->f[2]);
- OUT_RINGf (chan, color->f[3]);
+ BEGIN_NV04(push, NV50_3D(CLEAR_COLOR(0)), 4);
+ PUSH_DATAf(push, color->f[0]);
+ PUSH_DATAf(push, color->f[1]);
+ PUSH_DATAf(push, color->f[2]);
+ PUSH_DATAf(push, color->f[3]);
- if (MARK_RING(chan, 18, 2))
+ if (nouveau_pushbuf_space(push, 32 + sf->depth, 1, 0))
return;
- BEGIN_RING(chan, RING_3D(RT_CONTROL), 1);
- OUT_RING (chan, 1);
- BEGIN_RING(chan, RING_3D(RT_ADDRESS_HIGH(0)), 5);
- OUT_RELOCh(chan, bo, sf->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
- OUT_RELOCl(chan, bo, sf->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
- OUT_RING (chan, nv50_format_table[dst->format].rt);
- OUT_RING (chan, mt->level[sf->base.u.tex.level].tile_mode << 4);
- OUT_RING (chan, 0);
- BEGIN_RING(chan, RING_3D(RT_HORIZ(0)), 2);
- if (nouveau_bo_tile_layout(bo))
- OUT_RING(chan, sf->width);
+ PUSH_REFN(push, bo, mt->base.domain | NOUVEAU_BO_WR);
+
+ BEGIN_NV04(push, NV50_3D(RT_CONTROL), 1);
+ PUSH_DATA (push, 1);
+ BEGIN_NV04(push, NV50_3D(RT_ADDRESS_HIGH(0)), 5);
+ PUSH_DATAh(push, bo->offset + sf->offset);
+ PUSH_DATA (push, bo->offset + sf->offset);
+ PUSH_DATA (push, nv50_format_table[dst->format].rt);
+ PUSH_DATA (push, mt->level[sf->base.u.tex.level].tile_mode);
+ PUSH_DATA (push, 0);
+ BEGIN_NV04(push, NV50_3D(RT_HORIZ(0)), 2);
+ if (nouveau_bo_memtype(bo))
+ PUSH_DATA(push, sf->width);
else
- OUT_RING(chan, NV50_3D_RT_HORIZ_LINEAR | mt->level[0].pitch);
- OUT_RING (chan, sf->height);
- BEGIN_RING(chan, RING_3D(RT_ARRAY_MODE), 1);
- OUT_RING (chan, 1);
-
- if (!nouveau_bo_tile_layout(bo)) {
- BEGIN_RING(chan, RING_3D(ZETA_ENABLE), 1);
- OUT_RING (chan, 0);
+ PUSH_DATA(push, NV50_3D_RT_HORIZ_LINEAR | mt->level[0].pitch);
+ PUSH_DATA (push, sf->height);
+ BEGIN_NV04(push, NV50_3D(RT_ARRAY_MODE), 1);
+ PUSH_DATA (push, 1);
+
+ if (!nouveau_bo_memtype(bo)) {
+ BEGIN_NV04(push, NV50_3D(ZETA_ENABLE), 1);
+ PUSH_DATA (push, 0);
}
/* NOTE: only works with D3D clear flag (5097/0x143c bit 4) */
- BEGIN_RING(chan, RING_3D(VIEWPORT_HORIZ(0)), 2);
- OUT_RING (chan, (width << 16) | dstx);
- OUT_RING (chan, (height << 16) | dsty);
+ BEGIN_NV04(push, NV50_3D(VIEWPORT_HORIZ(0)), 2);
+ PUSH_DATA (push, (width << 16) | dstx);
+ PUSH_DATA (push, (height << 16) | dsty);
- BEGIN_RING(chan, RING_3D(CLEAR_BUFFERS), 1);
- OUT_RING (chan, 0x3c);
+ BEGIN_NI04(push, NV50_3D(CLEAR_BUFFERS), sf->depth);
+ for (z = 0; z < sf->depth; ++z) {
+ PUSH_DATA (push, 0x3c |
+ (z << NV50_3D_CLEAR_BUFFERS_LAYER__SHIFT));
+ }
nv50->dirty |= NV50_NEW_FRAMEBUFFER;
}
unsigned width, unsigned height)
{
struct nv50_context *nv50 = nv50_context(pipe);
- struct nv50_screen *screen = nv50->screen;
- struct nouveau_channel *chan = screen->base.channel;
+ struct nouveau_pushbuf *push = nv50->base.pushbuf;
struct nv50_miptree *mt = nv50_miptree(dst->texture);
struct nv50_surface *sf = nv50_surface(dst);
struct nouveau_bo *bo = mt->base.bo;
uint32_t mode = 0;
+ unsigned z;
- assert(nouveau_bo_tile_layout(bo)); /* ZETA cannot be linear */
+ assert(nouveau_bo_memtype(bo)); /* ZETA cannot be linear */
if (clear_flags & PIPE_CLEAR_DEPTH) {
- BEGIN_RING(chan, RING_3D(CLEAR_DEPTH), 1);
- OUT_RINGf (chan, depth);
+ BEGIN_NV04(push, NV50_3D(CLEAR_DEPTH), 1);
+ PUSH_DATAf(push, depth);
mode |= NV50_3D_CLEAR_BUFFERS_Z;
}
if (clear_flags & PIPE_CLEAR_STENCIL) {
- BEGIN_RING(chan, RING_3D(CLEAR_STENCIL), 1);
- OUT_RING (chan, stencil & 0xff);
+ BEGIN_NV04(push, NV50_3D(CLEAR_STENCIL), 1);
+ PUSH_DATA (push, stencil & 0xff);
mode |= NV50_3D_CLEAR_BUFFERS_S;
}
- if (MARK_RING(chan, 17, 2))
+ if (nouveau_pushbuf_space(push, 32 + sf->depth, 1, 0))
return;
- BEGIN_RING(chan, RING_3D(ZETA_ADDRESS_HIGH), 5);
- OUT_RELOCh(chan, bo, sf->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
- OUT_RELOCl(chan, bo, sf->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
- OUT_RING (chan, nv50_format_table[dst->format].rt);
- OUT_RING (chan, mt->level[sf->base.u.tex.level].tile_mode << 4);
- OUT_RING (chan, 0);
- BEGIN_RING(chan, RING_3D(ZETA_ENABLE), 1);
- OUT_RING (chan, 1);
- BEGIN_RING(chan, RING_3D(ZETA_HORIZ), 3);
- OUT_RING (chan, sf->width);
- OUT_RING (chan, sf->height);
- OUT_RING (chan, (1 << 16) | 1);
-
- BEGIN_RING(chan, RING_3D(VIEWPORT_HORIZ(0)), 2);
- OUT_RING (chan, (width << 16) | dstx);
- OUT_RING (chan, (height << 16) | dsty);
-
- BEGIN_RING(chan, RING_3D(CLEAR_BUFFERS), 1);
- OUT_RING (chan, mode);
+ PUSH_REFN(push, bo, mt->base.domain | NOUVEAU_BO_WR);
+
+ BEGIN_NV04(push, NV50_3D(ZETA_ADDRESS_HIGH), 5);
+ PUSH_DATAh(push, bo->offset + sf->offset);
+ PUSH_DATA (push, bo->offset + sf->offset);
+ PUSH_DATA (push, nv50_format_table[dst->format].rt);
+ PUSH_DATA (push, mt->level[sf->base.u.tex.level].tile_mode);
+ PUSH_DATA (push, 0);
+ BEGIN_NV04(push, NV50_3D(ZETA_ENABLE), 1);
+ PUSH_DATA (push, 1);
+ BEGIN_NV04(push, NV50_3D(ZETA_HORIZ), 3);
+ PUSH_DATA (push, sf->width);
+ PUSH_DATA (push, sf->height);
+ PUSH_DATA (push, (1 << 16) | 1);
+
+ BEGIN_NV04(push, NV50_3D(VIEWPORT_HORIZ(0)), 2);
+ PUSH_DATA (push, (width << 16) | dstx);
+ PUSH_DATA (push, (height << 16) | dsty);
+
+ BEGIN_NI04(push, NV50_3D(CLEAR_BUFFERS), sf->depth);
+ for (z = 0; z < sf->depth; ++z) {
+ PUSH_DATA (push, mode |
+ (z << NV50_3D_CLEAR_BUFFERS_LAYER__SHIFT));
+ }
nv50->dirty |= NV50_NEW_FRAMEBUFFER;
}
double depth, unsigned stencil)
{
struct nv50_context *nv50 = nv50_context(pipe);
- struct nouveau_channel *chan = nv50->screen->base.channel;
+ struct nouveau_pushbuf *push = nv50->base.pushbuf;
struct pipe_framebuffer_state *fb = &nv50->framebuffer;
unsigned i;
uint32_t mode = 0;
return;
if (buffers & PIPE_CLEAR_COLOR && fb->nr_cbufs) {
- BEGIN_RING(chan, RING_3D(CLEAR_COLOR(0)), 4);
- OUT_RINGf (chan, color->f[0]);
- OUT_RINGf (chan, color->f[1]);
- OUT_RINGf (chan, color->f[2]);
- OUT_RINGf (chan, color->f[3]);
+ BEGIN_NV04(push, NV50_3D(CLEAR_COLOR(0)), 4);
+ PUSH_DATAf(push, color->f[0]);
+ PUSH_DATAf(push, color->f[1]);
+ PUSH_DATAf(push, color->f[2]);
+ PUSH_DATAf(push, color->f[3]);
mode =
NV50_3D_CLEAR_BUFFERS_R | NV50_3D_CLEAR_BUFFERS_G |
NV50_3D_CLEAR_BUFFERS_B | NV50_3D_CLEAR_BUFFERS_A;
}
if (buffers & PIPE_CLEAR_DEPTH) {
- BEGIN_RING(chan, RING_3D(CLEAR_DEPTH), 1);
- OUT_RING (chan, fui(depth));
+ BEGIN_NV04(push, NV50_3D(CLEAR_DEPTH), 1);
+ PUSH_DATA (push, fui(depth));
mode |= NV50_3D_CLEAR_BUFFERS_Z;
}
if (buffers & PIPE_CLEAR_STENCIL) {
- BEGIN_RING(chan, RING_3D(CLEAR_STENCIL), 1);
- OUT_RING (chan, stencil & 0xff);
+ BEGIN_NV04(push, NV50_3D(CLEAR_STENCIL), 1);
+ PUSH_DATA (push, stencil & 0xff);
mode |= NV50_3D_CLEAR_BUFFERS_S;
}
- BEGIN_RING(chan, RING_3D(CLEAR_BUFFERS), 1);
- OUT_RING (chan, mode);
+ BEGIN_NV04(push, NV50_3D(CLEAR_BUFFERS), 1);
+ PUSH_DATA (push, mode);
for (i = 1; i < fb->nr_cbufs; i++) {
- BEGIN_RING(chan, RING_3D(CLEAR_BUFFERS), 1);
- OUT_RING (chan, (i << 6) | 0x3c);
+ BEGIN_NV04(push, NV50_3D(CLEAR_BUFFERS), 1);
+ PUSH_DATA (push, (i << 6) | 0x3c);
}
}
+/* =============================== BLIT CODE ===================================
+ */
+
+struct nv50_blitter
+{
+ struct nv50_program *fp[NV50_BLIT_MAX_TEXTURE_TYPES][NV50_BLIT_MODES];
+ struct nv50_program vp;
+
+ struct nv50_tsc_entry sampler[2]; /* nearest, bilinear */
+
+ pipe_mutex mutex;
+};
+
struct nv50_blitctx
{
- struct nv50_screen *screen;
+ struct nv50_context *nv50;
+ struct nv50_program *fp;
+ uint8_t mode;
+ uint16_t color_mask;
+ uint8_t filter;
+ enum pipe_texture_target target;
struct {
struct pipe_framebuffer_state fb;
struct nv50_program *vp;
struct nv50_program *fp;
unsigned num_textures[3];
unsigned num_samplers[3];
- struct pipe_sampler_view *texture;
- struct nv50_tsc_entry *sampler;
- unsigned dirty;
- unsigned clip_nr;
+ struct pipe_sampler_view *texture[2];
+ struct nv50_tsc_entry *sampler[2];
+ uint32_t dirty;
} saved;
- struct nv50_program vp;
- struct nv50_program fp;
- struct nv50_tsc_entry sampler[2]; /* nearest, bilinear */
- uint32_t fp_offset;
- uint16_t color_mask;
- uint8_t filter;
};
static void
-nv50_blitctx_make_vp(struct nv50_blitctx *blit)
+nv50_blitter_make_vp(struct nv50_blitter *blit)
{
static const uint32_t code[] =
{
- 0x10000001, /* mov b32 o[0x00] s[0x00] */ /* HPOS.x */
- 0x0423c788,
- 0x10000205, /* mov b32 o[0x04] s[0x04] */ /* HPOS.y */
- 0x0423c788,
- 0x10000409, /* mov b32 o[0x08] s[0x08] */ /* TEXC.x */
- 0x0423c788,
- 0x1000060d, /* mov b32 o[0x0c] s[0x0c] */ /* TEXC.y */
- 0x0423c788,
- 0x10000811, /* exit mov b32 o[0x10] s[0x10] */ /* TEXC.z */
- 0x0423c789,
+ 0x10000001, 0x0423c788, /* mov b32 o[0x00] s[0x00] */ /* HPOS.x */
+ 0x10000205, 0x0423c788, /* mov b32 o[0x04] s[0x04] */ /* HPOS.y */
+ 0x10000409, 0x0423c788, /* mov b32 o[0x08] s[0x08] */ /* TEXC.x */
+ 0x1000060d, 0x0423c788, /* mov b32 o[0x0c] s[0x0c] */ /* TEXC.y */
+ 0x10000811, 0x0423c789, /* mov b32 o[0x10] s[0x10] */ /* TEXC.z */
};
blit->vp.type = PIPE_SHADER_VERTEX;
blit->vp.out[1].hw = 2;
blit->vp.out[1].mask = 0x7;
blit->vp.out[1].sn = TGSI_SEMANTIC_GENERIC;
+ blit->vp.out[1].si = 8;
blit->vp.vp.attrs[0] = 0x73;
blit->vp.vp.psiz = 0x40;
blit->vp.vp.edgeflag = 0x40;
}
-static void
-nv50_blitctx_make_fp(struct nv50_blitctx *blit)
+void *
+nv50_blitter_make_fp(struct pipe_context *pipe,
+ unsigned mode,
+ enum pipe_texture_target ptarg)
{
- static const uint32_t code[] =
- {
- /* 3 coords RGBA in, RGBA out, also for Z32_FLOAT(_S8X24_UINT) */
- 0x80000000, /* interp $r0 v[0x0] */
- 0x80010004, /* interp $r1 v[0x4] */
- 0x80020009, /* interp $r2 flat v[0x8] */
- 0x00040780,
- 0xf6800001, /* texauto live { $r0,1,2,3 } $t0 $s0 { $r0,1,2 } */
- 0x0000c785, /* exit */
-
- /* 3 coords ZS in, S encoded in R, Z encoded in GBA (8_UNORM) */
- 0x80000000, /* interp $r0 v[0x00] */
- 0x80010004, /* interp $r1 v[0x04] */
- 0x80020009, /* interp $r2 flat v[0x8] */
- 0x00040780,
- 0xf6800001, /* texauto live { $r0,1,#,# } $t0 $s0 { $r0,1,2 } */
- 0x00000784,
- 0xc03f0009, /* mul f32 $r2 $r0 (2^24 - 1) */
- 0x04b7ffff,
- 0xa0000201, /* cvt f32 $r0 s32 $r1 */
- 0x44014780,
- 0xa0000409, /* cvt rni s32 $r2 f32 $r2 */
- 0x8c004780,
- 0xc0010001, /* mul f32 $r0 $r0 1/0xff */
- 0x03b8080b,
- 0xd03f0405, /* and b32 $r1 $r2 0x0000ff */
- 0x0000000f,
- 0xd000040d, /* and b32 $r3 $r2 0xff0000 */
- 0x000ff003,
- 0xd0000409, /* and b32 $r2 $r2 0x00ff00 */
- 0x00000ff3,
- 0xa0000205, /* cvt f32 $r1 s32 $r1 */
- 0x44014780,
- 0xa000060d, /* cvt f32 $r3 s32 $r3 */
- 0x44014780,
- 0xa0000409, /* cvt f32 $r2 s32 $r2 */
- 0x44014780,
- 0xc0010205, /* mul f32 $r1 $r1 1/0x0000ff */
- 0x03b8080b,
- 0xc001060d, /* mul f32 $r3 $r3 1/0x00ff00 */
- 0x0338080b,
- 0xc0010409, /* mul f32 $r2 $r2 1/0xff0000 */
- 0x0378080b,
- 0xf0000001, /* exit never nop */
- 0xe0000001,
-
- /* 3 coords ZS in, Z encoded in RGB, S encoded in A (U8_UNORM) */
- 0x80000000, /* interp $r0 v[0x00] */
- 0x80010004, /* interp $r1 v[0x04] */
- 0x80020009, /* interp $r2 flat v[0x8] */
- 0x00040780,
- 0xf6800001, /* texauto live { $r0,1,#,# } $t0 $s0 { $r0,1,2 } */
- 0x00000784,
- 0xc03f0009, /* mul f32 $r2 $r0 (2^24 - 1) */
- 0x04b7ffff,
- 0xa0000281, /* cvt f32 $r3 s32 $r1 */
- 0x44014780,
- 0xa0000409, /* cvt rni s32 $r2 f32 $r2 */
- 0x8c004780,
- 0xc001060d, /* mul f32 $r3 $r3 1/0xff */
- 0x03b8080b,
- 0xd03f0401, /* and b32 $r0 $r2 0x0000ff */
- 0x0000000f,
- 0xd0000405, /* and b32 $r1 $r2 0x00ff00 */
- 0x00000ff3,
- 0xd0000409, /* and b32 $r2 $r2 0xff0000 */
- 0x000ff003,
- 0xa0000001, /* cvt f32 $r0 s32 $r0 */
- 0x44014780,
- 0xa0000205, /* cvt f32 $r1 s32 $r1 */
- 0x44014780,
- 0xa0000409, /* cvt f32 $r2 s32 $r2 */
- 0x44014780,
- 0xc0010001, /* mul f32 $r0 $r0 1/0x0000ff */
- 0x03b8080b,
- 0xc0010205, /* mul f32 $r1 $r1 1/0x00ff00 */
- 0x0378080b,
- 0xc0010409, /* mul f32 $r2 $r2 1/0xff0000 */
- 0x0338080b,
- 0xf0000001, /* exit never nop */
- 0xe0000001
- };
+ struct ureg_program *ureg;
+ struct ureg_src tc;
+ struct ureg_dst out;
+ struct ureg_dst data;
+
+ const unsigned target = nv50_blit_get_tgsi_texture_target(ptarg);
+
+ boolean tex_rgbaz = FALSE;
+ boolean tex_s = FALSE;
+ boolean cvt_un8 = FALSE;
+
+ if (mode != NV50_BLIT_MODE_PASS &&
+ mode != NV50_BLIT_MODE_Z24X8 &&
+ mode != NV50_BLIT_MODE_X8Z24)
+ tex_s = TRUE;
+
+ if (mode != NV50_BLIT_MODE_X24S8 &&
+ mode != NV50_BLIT_MODE_S8X24 &&
+ mode != NV50_BLIT_MODE_XS)
+ tex_rgbaz = TRUE;
+
+ if (mode != NV50_BLIT_MODE_PASS &&
+ mode != NV50_BLIT_MODE_ZS &&
+ mode != NV50_BLIT_MODE_XS)
+ cvt_un8 = TRUE;
+
+ ureg = ureg_create(TGSI_PROCESSOR_FRAGMENT);
+ if (!ureg)
+ return NULL;
+
+ out = ureg_DECL_output(ureg, TGSI_SEMANTIC_COLOR, 0);
+ /* NOTE: use GENERIC[8] so we don't use the TEXCOORD slots on nvc0 */
+ tc = ureg_DECL_fs_input(
+ ureg, TGSI_SEMANTIC_GENERIC, 8, TGSI_INTERPOLATE_LINEAR);
+
+ data = ureg_DECL_temporary(ureg);
+
+ if (tex_s) {
+ ureg_TEX(ureg, ureg_writemask(data, TGSI_WRITEMASK_X),
+ target, tc, ureg_DECL_sampler(ureg, 1));
+ ureg_MOV(ureg, ureg_writemask(data, TGSI_WRITEMASK_Y),
+ ureg_scalar(ureg_src(data), TGSI_SWIZZLE_X));
+ }
+ if (tex_rgbaz) {
+ const unsigned mask = (mode == NV50_BLIT_MODE_PASS) ?
+ TGSI_WRITEMASK_XYZW : TGSI_WRITEMASK_X;
+ ureg_TEX(ureg, ureg_writemask(data, mask),
+ target, tc, ureg_DECL_sampler(ureg, 0));
+ }
+
+ if (cvt_un8) {
+ struct ureg_src mask;
+ struct ureg_src scale;
+ struct ureg_dst outz;
+ struct ureg_dst outs;
+ struct ureg_dst zdst3 = ureg_writemask(data, TGSI_WRITEMASK_XYZ);
+ struct ureg_dst zdst = ureg_writemask(data, TGSI_WRITEMASK_X);
+ struct ureg_dst sdst = ureg_writemask(data, TGSI_WRITEMASK_Y);
+ struct ureg_src zsrc3 = ureg_src(data);
+ struct ureg_src zsrc = ureg_scalar(zsrc3, TGSI_SWIZZLE_X);
+ struct ureg_src ssrc = ureg_scalar(zsrc3, TGSI_SWIZZLE_Y);
+ struct ureg_src zshuf;
+
+ mask = ureg_imm3u(ureg, 0x0000ff, 0x00ff00, 0xff0000);
+ scale = ureg_imm4f(ureg,
+ 1.0f / 0x0000ff, 1.0f / 0x00ff00, 1.0f / 0xff0000,
+ (1 << 24) - 1);
+
+ if (mode == NV50_BLIT_MODE_Z24S8 ||
+ mode == NV50_BLIT_MODE_X24S8 ||
+ mode == NV50_BLIT_MODE_Z24X8) {
+ outz = ureg_writemask(out, TGSI_WRITEMASK_XYZ);
+ outs = ureg_writemask(out, TGSI_WRITEMASK_W);
+ zshuf = ureg_src(data);
+ } else {
+ outz = ureg_writemask(out, TGSI_WRITEMASK_YZW);
+ outs = ureg_writemask(out, TGSI_WRITEMASK_X);
+ zshuf = ureg_swizzle(zsrc3, TGSI_SWIZZLE_W,
+ TGSI_SWIZZLE_X, TGSI_SWIZZLE_Y, TGSI_SWIZZLE_Z);
+ }
+
+ if (tex_s) {
+ ureg_I2F(ureg, sdst, ssrc);
+ ureg_MUL(ureg, outs, ssrc, ureg_scalar(scale, TGSI_SWIZZLE_X));
+ }
+
+ if (tex_rgbaz) {
+ ureg_MUL(ureg, zdst, zsrc, ureg_scalar(scale, TGSI_SWIZZLE_W));
+ ureg_F2I(ureg, zdst, zsrc);
+ ureg_AND(ureg, zdst3, zsrc, mask);
+ ureg_I2F(ureg, zdst3, zsrc3);
+ ureg_MUL(ureg, zdst3, zsrc3, scale);
+ ureg_MOV(ureg, outz, zshuf);
+ }
+ } else {
+ unsigned mask = TGSI_WRITEMASK_XYZW;
+
+ if (mode != NV50_BLIT_MODE_PASS) {
+ mask &= ~TGSI_WRITEMASK_ZW;
+ if (!tex_s)
+ mask = TGSI_WRITEMASK_X;
+ if (!tex_rgbaz)
+ mask = TGSI_WRITEMASK_Y;
+ }
+ ureg_MOV(ureg, ureg_writemask(out, mask), ureg_src(data));
+ }
+ ureg_END(ureg);
- blit->fp.type = PIPE_SHADER_FRAGMENT;
- blit->fp.translated = TRUE;
- blit->fp.code = (uint32_t *)code; /* const_cast */
- blit->fp.code_size = sizeof(code);
- blit->fp.max_gpr = 4;
- blit->fp.max_out = 4;
- blit->fp.in_nr = 1;
- blit->fp.in[0].mask = 0x7; /* last component flat */
- blit->fp.in[0].linear = 1;
- blit->fp.in[0].sn = TGSI_SEMANTIC_GENERIC;
- blit->fp.out_nr = 1;
- blit->fp.out[0].mask = 0xf;
- blit->fp.out[0].sn = TGSI_SEMANTIC_COLOR;
- blit->fp.fp.interp = 0x00020403;
- blit->fp.gp.primid = 0x80;
+ return ureg_create_shader_and_destroy(ureg, pipe);
}
static void
-nv50_blitctx_make_sampler(struct nv50_blitctx *blit)
+nv50_blitter_make_sampler(struct nv50_blitter *blit)
{
/* clamp to edge, min/max lod = 0, nearest filtering */
blit->sampler[0].id = -1;
- blit->sampler[0].tsc[0] = 0x00000092;
- blit->sampler[0].tsc[1] = 0x00000051;
+ blit->sampler[0].tsc[0] = NV50_TSC_0_SRGB_CONVERSION_ALLOWED |
+ (NV50_TSC_WRAP_CLAMP_TO_EDGE << NV50_TSC_0_WRAPS__SHIFT) |
+ (NV50_TSC_WRAP_CLAMP_TO_EDGE << NV50_TSC_0_WRAPT__SHIFT) |
+ (NV50_TSC_WRAP_CLAMP_TO_EDGE << NV50_TSC_0_WRAPR__SHIFT);
+ blit->sampler[0].tsc[1] =
+ NV50_TSC_1_MAGF_NEAREST | NV50_TSC_1_MINF_NEAREST | NV50_TSC_1_MIPF_NONE;
/* clamp to edge, min/max lod = 0, bilinear filtering */
blit->sampler[1].id = -1;
- blit->sampler[1].tsc[0] = 0x00000092;
- blit->sampler[1].tsc[1] = 0x00000062;
+ blit->sampler[1].tsc[0] = blit->sampler[0].tsc[0];
+ blit->sampler[1].tsc[1] =
+ NV50_TSC_1_MAGF_LINEAR | NV50_TSC_1_MINF_LINEAR | NV50_TSC_1_MIPF_NONE;
}
-/* Since shaders cannot export stencil, we cannot copy stencil values when
- * rendering to ZETA, so we attach the ZS surface to a colour render target.
- */
-static INLINE enum pipe_format
-nv50_blit_zeta_to_colour_format(enum pipe_format format)
+unsigned
+nv50_blit_select_mode(const struct pipe_blit_info *info)
{
- switch (format) {
- case PIPE_FORMAT_Z16_UNORM: return PIPE_FORMAT_R16_UNORM;
+ const unsigned mask = info->mask;
+
+ switch (info->dst.resource->format) {
case PIPE_FORMAT_Z24_UNORM_S8_UINT:
+ case PIPE_FORMAT_Z24X8_UNORM:
+ switch (mask & PIPE_MASK_ZS) {
+ case PIPE_MASK_ZS: return NV50_BLIT_MODE_Z24S8;
+ case PIPE_MASK_Z: return NV50_BLIT_MODE_Z24X8;
+ default:
+ return NV50_BLIT_MODE_X24S8;
+ }
case PIPE_FORMAT_S8_UINT_Z24_UNORM:
- case PIPE_FORMAT_Z24X8_UNORM: return PIPE_FORMAT_R8G8B8A8_UNORM;
- case PIPE_FORMAT_Z32_FLOAT: return PIPE_FORMAT_R32_FLOAT;
- case PIPE_FORMAT_Z32_FLOAT_S8X24_UINT: return PIPE_FORMAT_R32G32_FLOAT;
+ switch (mask & PIPE_MASK_ZS) {
+ case PIPE_MASK_ZS: return NV50_BLIT_MODE_S8Z24;
+ case PIPE_MASK_Z: return NV50_BLIT_MODE_X8Z24;
+ default:
+ return NV50_BLIT_MODE_S8X24;
+ }
+ case PIPE_FORMAT_Z32_FLOAT:
+ case PIPE_FORMAT_Z32_FLOAT_S8X24_UINT:
+ switch (mask & PIPE_MASK_ZS) {
+ case PIPE_MASK_ZS: return NV50_BLIT_MODE_ZS;
+ case PIPE_MASK_Z: return NV50_BLIT_MODE_PASS;
+ default:
+ return NV50_BLIT_MODE_XS;
+ }
default:
- assert(0);
- return PIPE_FORMAT_NONE;
+ return NV50_BLIT_MODE_PASS;
}
}
static void
-nv50_blitctx_get_color_mask_and_fp(struct nv50_blitctx *blit,
- enum pipe_format format, uint8_t mask)
+nv50_blit_select_fp(struct nv50_blitctx *ctx, const struct pipe_blit_info *info)
{
- blit->color_mask = 0;
+ struct nv50_blitter *blitter = ctx->nv50->screen->blitter;
- switch (format) {
- case PIPE_FORMAT_Z24X8_UNORM:
- case PIPE_FORMAT_Z24_UNORM_S8_UINT:
- blit->fp_offset = 160;
- if (mask & PIPE_MASK_Z)
- blit->color_mask |= 0x0111;
- if (mask & PIPE_MASK_S)
- blit->color_mask |= 0x1000;
- break;
- case PIPE_FORMAT_S8_UINT_Z24_UNORM:
- blit->fp_offset = 24;
- if (mask & PIPE_MASK_Z)
- blit->color_mask |= 0x1110;
- if (mask & PIPE_MASK_S)
- blit->color_mask |= 0x0001;
- break;
- default:
- blit->fp_offset = 0;
- if (mask & (PIPE_MASK_R | PIPE_MASK_Z)) blit->color_mask |= 0x0001;
- if (mask & (PIPE_MASK_G | PIPE_MASK_S)) blit->color_mask |= 0x0010;
- if (mask & PIPE_MASK_B) blit->color_mask |= 0x0100;
- if (mask & PIPE_MASK_A) blit->color_mask |= 0x1000;
- break;
+ const enum pipe_texture_target ptarg =
+ nv50_blit_reinterpret_pipe_texture_target(info->src.resource->target);
+
+ const unsigned targ = nv50_blit_texture_type(ptarg);
+ const unsigned mode = ctx->mode;
+
+ if (!blitter->fp[targ][mode]) {
+ pipe_mutex_lock(blitter->mutex);
+ if (!blitter->fp[targ][mode])
+ blitter->fp[targ][mode] =
+ nv50_blitter_make_fp(&ctx->nv50->base.pipe, mode, ptarg);
+ pipe_mutex_unlock(blitter->mutex);
}
+ ctx->fp = blitter->fp[targ][mode];
}
static void
-nv50_blit_set_dst(struct nv50_context *nv50,
- struct pipe_resource *res, unsigned level, unsigned layer)
+nv50_blit_set_dst(struct nv50_blitctx *ctx,
+ struct pipe_resource *res, unsigned level, unsigned layer,
+ enum pipe_format format)
{
+ struct nv50_context *nv50 = ctx->nv50;
struct pipe_context *pipe = &nv50->base.pipe;
struct pipe_surface templ;
- if (util_format_is_depth_or_stencil(res->format))
- templ.format = nv50_blit_zeta_to_colour_format(res->format);
+ if (util_format_is_depth_or_stencil(format))
+ templ.format = nv50_blit_zeta_to_colour_format(format);
else
- templ.format = res->format;
+ templ.format = format;
- templ.usage = PIPE_USAGE_STREAM;
templ.u.tex.level = level;
templ.u.tex.first_layer = templ.u.tex.last_layer = layer;
+ if (layer == -1) {
+ templ.u.tex.first_layer = 0;
+ templ.u.tex.last_layer =
+ (res->target == PIPE_TEXTURE_3D ? res->depth0 : res->array_size) - 1;
+ }
+
nv50->framebuffer.cbufs[0] = nv50_miptree_surface_new(pipe, res, &templ);
nv50->framebuffer.nr_cbufs = 1;
nv50->framebuffer.zsbuf = NULL;
nv50->framebuffer.height = nv50->framebuffer.cbufs[0]->height;
}
-static INLINE void
-nv50_blit_fixup_tic_entry(struct pipe_sampler_view *view)
-{
- struct nv50_tic_entry *ent = nv50_tic_entry(view);
-
- ent->tic[2] &= ~(1 << 31); /* scaled coordinates, ok with 3d textures ? */
-
- /* magic: */
-
- ent->tic[3] = 0x20000000; /* affects quality of near vertical edges in MS8 */
-}
-
static void
-nv50_blit_set_src(struct nv50_context *nv50,
- struct pipe_resource *res, unsigned level, unsigned layer)
+nv50_blit_set_src(struct nv50_blitctx *blit,
+ struct pipe_resource *res, unsigned level, unsigned layer,
+ enum pipe_format format, const uint8_t filter)
{
+ struct nv50_context *nv50 = blit->nv50;
struct pipe_context *pipe = &nv50->base.pipe;
struct pipe_sampler_view templ;
+ uint32_t flags;
+ enum pipe_texture_target target;
- templ.format = res->format;
- templ.u.tex.first_layer = templ.u.tex.last_layer = layer;
+ target = nv50_blit_reinterpret_pipe_texture_target(res->target);
+
+ templ.format = format;
templ.u.tex.first_level = templ.u.tex.last_level = level;
+ templ.u.tex.first_layer = templ.u.tex.last_layer = layer;
templ.swizzle_r = PIPE_SWIZZLE_RED;
templ.swizzle_g = PIPE_SWIZZLE_GREEN;
templ.swizzle_b = PIPE_SWIZZLE_BLUE;
templ.swizzle_a = PIPE_SWIZZLE_ALPHA;
- nv50->textures[2][0] = nv50_create_sampler_view(pipe, res, &templ);
+ if (layer == -1) {
+ templ.u.tex.first_layer = 0;
+ templ.u.tex.last_layer =
+ (res->target == PIPE_TEXTURE_3D ? res->depth0 : res->array_size) - 1;
+ }
+
+ flags = res->last_level ? 0 : NV50_TEXVIEW_SCALED_COORDS;
+ if (filter && res->nr_samples == 8)
+ flags |= NV50_TEXVIEW_FILTER_MSAA8;
- nv50_blit_fixup_tic_entry(nv50->textures[2][0]);
+ nv50->textures[2][0] = nv50_create_texture_view(
+ pipe, res, &templ, flags, target);
+ nv50->textures[2][1] = NULL;
nv50->num_textures[0] = nv50->num_textures[1] = 0;
nv50->num_textures[2] = 1;
+
+ templ.format = nv50_zs_to_s_format(format);
+ if (templ.format != res->format) {
+ nv50->textures[2][1] = nv50_create_texture_view(
+ pipe, res, &templ, flags, target);
+ nv50->num_textures[2] = 2;
+ }
}
static void
nv50_blitctx_prepare_state(struct nv50_blitctx *blit)
{
- struct nouveau_channel *chan = blit->screen->base.channel;
+ struct nouveau_pushbuf *push = blit->nv50->base.pushbuf;
+
+ if (blit->nv50->cond_query) {
+ BEGIN_NV04(push, NV50_3D(COND_MODE), 1);
+ PUSH_DATA (push, NV50_3D_COND_MODE_ALWAYS);
+ }
/* blend state */
- BEGIN_RING(chan, RING_3D(COLOR_MASK(0)), 1);
- OUT_RING (chan, blit->color_mask);
- BEGIN_RING(chan, RING_3D(BLEND_ENABLE(0)), 1);
- OUT_RING (chan, 0);
- BEGIN_RING(chan, RING_3D(LOGIC_OP_ENABLE), 1);
- OUT_RING (chan, 0);
+ BEGIN_NV04(push, NV50_3D(COLOR_MASK(0)), 1);
+ PUSH_DATA (push, blit->color_mask);
+ BEGIN_NV04(push, NV50_3D(BLEND_ENABLE(0)), 1);
+ PUSH_DATA (push, 0);
+ BEGIN_NV04(push, NV50_3D(LOGIC_OP_ENABLE), 1);
+ PUSH_DATA (push, 0);
/* rasterizer state */
#ifndef NV50_SCISSORS_CLIPPING
- BEGIN_RING(chan, RING_3D(SCISSOR_ENABLE(0)), 1);
- OUT_RING (chan, 1);
+ BEGIN_NV04(push, NV50_3D(SCISSOR_ENABLE(0)), 1);
+ PUSH_DATA (push, 1);
#endif
- BEGIN_RING(chan, RING_3D(VERTEX_TWO_SIDE_ENABLE), 1);
- OUT_RING (chan, 0);
- BEGIN_RING(chan, RING_3D(FRAG_COLOR_CLAMP_EN), 1);
- OUT_RING (chan, 0);
- BEGIN_RING(chan, RING_3D(MULTISAMPLE_ENABLE), 1);
- OUT_RING (chan, 0);
- BEGIN_RING(chan, RING_3D(MSAA_MASK(0)), 4);
- OUT_RING (chan, 0xffff);
- OUT_RING (chan, 0xffff);
- OUT_RING (chan, 0xffff);
- OUT_RING (chan, 0xffff);
- BEGIN_RING(chan, RING_3D(POLYGON_MODE_FRONT), 3);
- OUT_RING (chan, NV50_3D_POLYGON_MODE_FRONT_FILL);
- OUT_RING (chan, NV50_3D_POLYGON_MODE_BACK_FILL);
- OUT_RING (chan, 0);
- BEGIN_RING(chan, RING_3D(CULL_FACE_ENABLE), 1);
- OUT_RING (chan, 0);
- BEGIN_RING(chan, RING_3D(POLYGON_STIPPLE_ENABLE), 1);
- OUT_RING (chan, 0);
- BEGIN_RING(chan, RING_3D(POLYGON_OFFSET_FILL_ENABLE), 1);
- OUT_RING (chan, 0);
+ BEGIN_NV04(push, NV50_3D(VERTEX_TWO_SIDE_ENABLE), 1);
+ PUSH_DATA (push, 0);
+ BEGIN_NV04(push, NV50_3D(FRAG_COLOR_CLAMP_EN), 1);
+ PUSH_DATA (push, 0);
+ BEGIN_NV04(push, NV50_3D(MULTISAMPLE_ENABLE), 1);
+ PUSH_DATA (push, 0);
+ BEGIN_NV04(push, NV50_3D(MSAA_MASK(0)), 4);
+ PUSH_DATA (push, 0xffff);
+ PUSH_DATA (push, 0xffff);
+ PUSH_DATA (push, 0xffff);
+ PUSH_DATA (push, 0xffff);
+ BEGIN_NV04(push, NV50_3D(POLYGON_MODE_FRONT), 3);
+ PUSH_DATA (push, NV50_3D_POLYGON_MODE_FRONT_FILL);
+ PUSH_DATA (push, NV50_3D_POLYGON_MODE_BACK_FILL);
+ PUSH_DATA (push, 0);
+ BEGIN_NV04(push, NV50_3D(CULL_FACE_ENABLE), 1);
+ PUSH_DATA (push, 0);
+ BEGIN_NV04(push, NV50_3D(POLYGON_STIPPLE_ENABLE), 1);
+ PUSH_DATA (push, 0);
+ BEGIN_NV04(push, NV50_3D(POLYGON_OFFSET_FILL_ENABLE), 1);
+ PUSH_DATA (push, 0);
/* zsa state */
- BEGIN_RING(chan, RING_3D(DEPTH_TEST_ENABLE), 1);
- OUT_RING (chan, 0);
- BEGIN_RING(chan, RING_3D(STENCIL_ENABLE), 1);
- OUT_RING (chan, 0);
- BEGIN_RING(chan, RING_3D(ALPHA_TEST_ENABLE), 1);
- OUT_RING (chan, 0);
+ BEGIN_NV04(push, NV50_3D(DEPTH_TEST_ENABLE), 1);
+ PUSH_DATA (push, 0);
+ BEGIN_NV04(push, NV50_3D(STENCIL_ENABLE), 1);
+ PUSH_DATA (push, 0);
+ BEGIN_NV04(push, NV50_3D(ALPHA_TEST_ENABLE), 1);
+ PUSH_DATA (push, 0);
}
static void
-nv50_blitctx_pre_blit(struct nv50_blitctx *blit, struct nv50_context *nv50)
+nv50_blitctx_pre_blit(struct nv50_blitctx *ctx)
{
+ struct nv50_context *nv50 = ctx->nv50;
+ struct nv50_blitter *blitter = nv50->screen->blitter;
int s;
- blit->saved.fb.width = nv50->framebuffer.width;
- blit->saved.fb.height = nv50->framebuffer.height;
- blit->saved.fb.nr_cbufs = nv50->framebuffer.nr_cbufs;
- blit->saved.fb.cbufs[0] = nv50->framebuffer.cbufs[0];
- blit->saved.fb.zsbuf = nv50->framebuffer.zsbuf;
+ ctx->saved.fb.width = nv50->framebuffer.width;
+ ctx->saved.fb.height = nv50->framebuffer.height;
+ ctx->saved.fb.nr_cbufs = nv50->framebuffer.nr_cbufs;
+ ctx->saved.fb.cbufs[0] = nv50->framebuffer.cbufs[0];
+ ctx->saved.fb.zsbuf = nv50->framebuffer.zsbuf;
- blit->saved.vp = nv50->vertprog;
- blit->saved.gp = nv50->gmtyprog;
- blit->saved.fp = nv50->fragprog;
+ ctx->saved.vp = nv50->vertprog;
+ ctx->saved.gp = nv50->gmtyprog;
+ ctx->saved.fp = nv50->fragprog;
- nv50->vertprog = &blit->vp;
+ nv50->vertprog = &blitter->vp;
nv50->gmtyprog = NULL;
- nv50->fragprog = &blit->fp;
-
- blit->saved.clip_nr = nv50->clip.nr;
-
- nv50->clip.nr = 0;
+ nv50->fragprog = ctx->fp;
for (s = 0; s < 3; ++s) {
- blit->saved.num_textures[s] = nv50->num_textures[s];
- blit->saved.num_samplers[s] = nv50->num_samplers[s];
+ ctx->saved.num_textures[s] = nv50->num_textures[s];
+ ctx->saved.num_samplers[s] = nv50->num_samplers[s];
}
- blit->saved.texture = nv50->textures[2][0];
- blit->saved.sampler = nv50->samplers[2][0];
+ ctx->saved.texture[0] = nv50->textures[2][0];
+ ctx->saved.texture[1] = nv50->textures[2][1];
+ ctx->saved.sampler[0] = nv50->samplers[2][0];
+ ctx->saved.sampler[1] = nv50->samplers[2][1];
- nv50->samplers[2][0] = &blit->sampler[blit->filter];
+ nv50->samplers[2][0] = &blitter->sampler[ctx->filter];
+ nv50->samplers[2][1] = &blitter->sampler[ctx->filter];
nv50->num_samplers[0] = nv50->num_samplers[1] = 0;
- nv50->num_samplers[2] = 1;
+ nv50->num_samplers[2] = 2;
+
+ ctx->saved.dirty = nv50->dirty;
- blit->saved.dirty = nv50->dirty;
+ nouveau_bufctx_reset(nv50->bufctx_3d, NV50_BIND_FB);
+ nouveau_bufctx_reset(nv50->bufctx_3d, NV50_BIND_TEXTURES);
nv50->dirty =
NV50_NEW_FRAMEBUFFER |
}
static void
-nv50_blitctx_post_blit(struct nv50_context *nv50, struct nv50_blitctx *blit)
+nv50_blitctx_post_blit(struct nv50_blitctx *blit)
{
+ struct nv50_context *nv50 = blit->nv50;
int s;
pipe_surface_reference(&nv50->framebuffer.cbufs[0], NULL);
nv50->gmtyprog = blit->saved.gp;
nv50->fragprog = blit->saved.fp;
- nv50->clip.nr = blit->saved.clip_nr;
-
pipe_sampler_view_reference(&nv50->textures[2][0], NULL);
+ pipe_sampler_view_reference(&nv50->textures[2][1], NULL);
for (s = 0; s < 3; ++s) {
nv50->num_textures[s] = blit->saved.num_textures[s];
nv50->num_samplers[s] = blit->saved.num_samplers[s];
}
- nv50->textures[2][0] = blit->saved.texture;
- nv50->samplers[2][0] = blit->saved.sampler;
+ nv50->textures[2][0] = blit->saved.texture[0];
+ nv50->textures[2][1] = blit->saved.texture[1];
+ nv50->samplers[2][0] = blit->saved.sampler[0];
+ nv50->samplers[2][1] = blit->saved.sampler[1];
+
+ if (nv50->cond_query)
+ nv50->base.pipe.render_condition(&nv50->base.pipe, nv50->cond_query,
+ nv50->cond_mode);
+
+ nouveau_bufctx_reset(nv50->bufctx_3d, NV50_BIND_FB);
+ nouveau_bufctx_reset(nv50->bufctx_3d, NV50_BIND_TEXTURES);
nv50->dirty = blit->saved.dirty |
(NV50_NEW_FRAMEBUFFER | NV50_NEW_SCISSOR | NV50_NEW_SAMPLE_MASK |
NV50_NEW_VERTPROG | NV50_NEW_GMTYPROG | NV50_NEW_FRAGPROG);
}
+
static void
-nv50_resource_resolve(struct pipe_context *pipe,
- const struct pipe_resolve_info *info)
+nv50_blit_3d(struct nv50_context *nv50, const struct pipe_blit_info *info)
{
- struct nv50_context *nv50 = nv50_context(pipe);
- struct nv50_screen *screen = nv50->screen;
- struct nv50_blitctx *blit = screen->blitctx;
- struct nouveau_channel *chan = screen->base.channel;
- struct pipe_resource *src = info->src.res;
- struct pipe_resource *dst = info->dst.res;
+ struct nv50_blitctx *blit = nv50->blit;
+ struct nouveau_pushbuf *push = nv50->base.pushbuf;
+ struct pipe_resource *src = info->src.resource;
+ struct pipe_resource *dst = info->dst.resource;
+ int32_t minx, maxx, miny, maxy;
+ int32_t i;
float x0, x1, y0, y1, z;
+ float dz;
float x_range, y_range;
- nv50_blitctx_get_color_mask_and_fp(blit, dst->format, info->mask);
+ blit->mode = nv50_blit_select_mode(info);
+ blit->color_mask = nv50_blit_derive_color_mask(info);
+ blit->filter = nv50_blit_get_filter(info);
- blit->filter = util_format_is_depth_or_stencil(dst->format) ? 0 : 1;
+ nv50_blit_select_fp(blit, info);
+ nv50_blitctx_pre_blit(blit);
- nv50_blitctx_pre_blit(blit, nv50);
-
- nv50_blit_set_dst(nv50, dst, info->dst.level, info->dst.layer);
- nv50_blit_set_src(nv50, src, 0, info->src.layer);
+ nv50_blit_set_dst(blit, dst, info->dst.level, 0, info->dst.format);
+ nv50_blit_set_src(blit, src, info->src.level, -1, info->src.format,
+ blit->filter);
nv50_blitctx_prepare_state(blit);
nv50_state_validate(nv50, ~0, 36);
- x_range =
- (float)(info->src.x1 - info->src.x0) /
- (float)(info->dst.x1 - info->dst.x0);
- y_range =
- (float)(info->src.y1 - info->src.y0) /
- (float)(info->dst.y1 - info->dst.y0);
+ x_range = (float)info->src.box.width / (float)info->dst.box.width;
+ y_range = (float)info->src.box.height / (float)info->dst.box.height;
- x0 = (float)info->src.x0 - x_range * (float)info->dst.x0;
- y0 = (float)info->src.y0 - y_range * (float)info->dst.y0;
+ x0 = (float)info->src.box.x - x_range * (float)info->dst.box.x;
+ y0 = (float)info->src.box.y - y_range * (float)info->dst.box.y;
x1 = x0 + 16384.0f * x_range;
y1 = y0 + 16384.0f * y_range;
y0 *= (float)(1 << nv50_miptree(src)->ms_y);
y1 *= (float)(1 << nv50_miptree(src)->ms_y);
- z = (float)info->src.layer;
+ if (src->last_level > 0) {
+ /* If there are mip maps, GPU always assumes normalized coordinates. */
+ const unsigned l = info->src.level;
+ const float fh = u_minify(src->width0 << nv50_miptree(src)->ms_x, l);
+ const float fv = u_minify(src->height0 << nv50_miptree(src)->ms_y, l);
+ x0 /= fh;
+ x1 /= fh;
+ y0 /= fv;
+ y1 /= fv;
+ }
- BEGIN_RING(chan, RING_3D(FP_START_ID), 1);
- OUT_RING (chan,
- blit->fp.code_base + blit->fp_offset);
+ /* XXX: multiply by 6 for cube arrays ? */
+ dz = (float)info->src.box.depth / (float)info->dst.box.depth;
+ z = (float)info->src.box.z;
+ if (nv50_miptree(src)->layout_3d)
+ z += 0.5f * dz;
- BEGIN_RING(chan, RING_3D(VIEWPORT_TRANSFORM_EN), 1);
- OUT_RING (chan, 0);
+ BEGIN_NV04(push, NV50_3D(VIEWPORT_TRANSFORM_EN), 1);
+ PUSH_DATA (push, 0);
/* Draw a large triangle in screen coordinates covering the whole
* render target, with scissors defining the destination region.
* arranged in a way to yield the desired offset and scale.
*/
- BEGIN_RING(chan, RING_3D(SCISSOR_HORIZ(0)), 2);
- OUT_RING (chan, (info->dst.x1 << 16) | info->dst.x0);
- OUT_RING (chan, (info->dst.y1 << 16) | info->dst.y0);
-
- BEGIN_RING(chan, RING_3D(VERTEX_BEGIN_GL), 1);
- OUT_RING (chan, NV50_3D_VERTEX_BEGIN_GL_PRIMITIVE_TRIANGLES);
- BEGIN_RING(chan, RING_3D(VTX_ATTR_3F_X(1)), 3);
- OUT_RINGf (chan, x0);
- OUT_RINGf (chan, y0);
- OUT_RINGf (chan, z);
- BEGIN_RING(chan, RING_3D(VTX_ATTR_2F_X(0)), 2);
- OUT_RINGf (chan, 0.0f);
- OUT_RINGf (chan, 0.0f);
- BEGIN_RING(chan, RING_3D(VTX_ATTR_3F_X(1)), 3);
- OUT_RINGf (chan, x1);
- OUT_RINGf (chan, y0);
- OUT_RINGf (chan, z);
- BEGIN_RING(chan, RING_3D(VTX_ATTR_2F_X(0)), 2);
- OUT_RINGf (chan, 16384 << nv50_miptree(dst)->ms_x);
- OUT_RINGf (chan, 0.0f);
- BEGIN_RING(chan, RING_3D(VTX_ATTR_3F_X(1)), 3);
- OUT_RINGf (chan, x0);
- OUT_RINGf (chan, y1);
- OUT_RINGf (chan, z);
- BEGIN_RING(chan, RING_3D(VTX_ATTR_2F_X(0)), 2);
- OUT_RINGf (chan, 0.0f);
- OUT_RINGf (chan, 16384 << nv50_miptree(dst)->ms_y);
- BEGIN_RING(chan, RING_3D(VERTEX_END_GL), 1);
- OUT_RING (chan, 0);
+ minx = info->dst.box.x;
+ maxx = info->dst.box.x + info->dst.box.width;
+ miny = info->dst.box.y;
+ maxy = info->dst.box.y + info->dst.box.height;
+ if (info->scissor_enable) {
+ minx = MAX2(minx, info->scissor.minx);
+ maxx = MIN2(maxx, info->scissor.maxx);
+ miny = MAX2(miny, info->scissor.miny);
+ maxy = MIN2(maxy, info->scissor.maxy);
+ }
+ BEGIN_NV04(push, NV50_3D(SCISSOR_HORIZ(0)), 2);
+ PUSH_DATA (push, (maxx << 16) | minx);
+ PUSH_DATA (push, (maxy << 16) | miny);
+
+ for (i = 0; i < info->dst.box.depth; ++i, z += dz) {
+ if (info->dst.box.z + i) {
+ BEGIN_NV04(push, NV50_3D(LAYER), 1);
+ PUSH_DATA (push, info->dst.box.z + i);
+ }
+ PUSH_SPACE(push, 32);
+ BEGIN_NV04(push, NV50_3D(VERTEX_BEGIN_GL), 1);
+ PUSH_DATA (push, NV50_3D_VERTEX_BEGIN_GL_PRIMITIVE_TRIANGLES);
+ BEGIN_NV04(push, NV50_3D(VTX_ATTR_3F_X(1)), 3);
+ PUSH_DATAf(push, x0);
+ PUSH_DATAf(push, y0);
+ PUSH_DATAf(push, z);
+ BEGIN_NV04(push, NV50_3D(VTX_ATTR_2F_X(0)), 2);
+ PUSH_DATAf(push, 0.0f);
+ PUSH_DATAf(push, 0.0f);
+ BEGIN_NV04(push, NV50_3D(VTX_ATTR_3F_X(1)), 3);
+ PUSH_DATAf(push, x1);
+ PUSH_DATAf(push, y0);
+ PUSH_DATAf(push, z);
+ BEGIN_NV04(push, NV50_3D(VTX_ATTR_2F_X(0)), 2);
+ PUSH_DATAf(push, 16384 << nv50_miptree(dst)->ms_x);
+ PUSH_DATAf(push, 0.0f);
+ BEGIN_NV04(push, NV50_3D(VTX_ATTR_3F_X(1)), 3);
+ PUSH_DATAf(push, x0);
+ PUSH_DATAf(push, y1);
+ PUSH_DATAf(push, z);
+ BEGIN_NV04(push, NV50_3D(VTX_ATTR_2F_X(0)), 2);
+ PUSH_DATAf(push, 0.0f);
+ PUSH_DATAf(push, 16384 << nv50_miptree(dst)->ms_y);
+ BEGIN_NV04(push, NV50_3D(VERTEX_END_GL), 1);
+ PUSH_DATA (push, 0);
+ }
+ if (info->dst.box.z + info->dst.box.depth - 1) {
+ BEGIN_NV04(push, NV50_3D(LAYER), 1);
+ PUSH_DATA (push, 0);
+ }
/* re-enable normally constant state */
- BEGIN_RING(chan, RING_3D(VIEWPORT_TRANSFORM_EN), 1);
- OUT_RING (chan, 1);
+ BEGIN_NV04(push, NV50_3D(VIEWPORT_TRANSFORM_EN), 1);
+ PUSH_DATA (push, 1);
- nv50_blitctx_post_blit(nv50, blit);
+ nv50_blitctx_post_blit(blit);
+}
+
+static void
+nv50_blit_eng2d(struct nv50_context *nv50, const struct pipe_blit_info *info)
+{
+ struct nouveau_pushbuf *push = nv50->base.pushbuf;
+ struct nv50_miptree *dst = nv50_miptree(info->dst.resource);
+ struct nv50_miptree *src = nv50_miptree(info->src.resource);
+ const int32_t srcx_adj = info->src.box.width < 0 ? -1 : 0;
+ const int32_t srcy_adj = info->src.box.height < 0 ? -1 : 0;
+ const int32_t dz = info->dst.box.z;
+ const int32_t sz = info->src.box.z;
+ uint32_t dstw, dsth;
+ int32_t dstx, dsty;
+ int64_t srcx, srcy;
+ int64_t du_dx, dv_dy;
+ int i;
+ uint32_t mode;
+ const uint32_t mask = nv50_blit_eng2d_get_mask(info);
+
+ mode = nv50_blit_get_filter(info) ?
+ NV50_2D_BLIT_CONTROL_FILTER_BILINEAR :
+ NV50_2D_BLIT_CONTROL_FILTER_POINT_SAMPLE;
+ mode |= (src->base.base.nr_samples > dst->base.base.nr_samples) ?
+ NV50_2D_BLIT_CONTROL_ORIGIN_CORNER : NV50_2D_BLIT_CONTROL_ORIGIN_CENTER;
+
+ du_dx = ((int64_t)info->src.box.width << 32) / info->dst.box.width;
+ dv_dy = ((int64_t)info->src.box.height << 32) / info->dst.box.height;
+
+ nv50_2d_texture_set(push, 1, dst, info->dst.level, dz, info->dst.format);
+ nv50_2d_texture_set(push, 0, src, info->src.level, sz, info->src.format);
+
+ if (info->scissor_enable) {
+ BEGIN_NV04(push, NV50_2D(CLIP_X), 5);
+ PUSH_DATA (push, info->scissor.minx << dst->ms_x);
+ PUSH_DATA (push, info->scissor.miny << dst->ms_y);
+ PUSH_DATA (push, (info->scissor.maxx - info->scissor.minx) << dst->ms_x);
+ PUSH_DATA (push, (info->scissor.maxy - info->scissor.miny) << dst->ms_y);
+ PUSH_DATA (push, 1); /* enable */
+ }
+
+ if (mask != 0xffffffff) {
+ BEGIN_NV04(push, NV50_2D(ROP), 1);
+ PUSH_DATA (push, 0xca); /* DPSDxax */
+ BEGIN_NV04(push, NV50_2D(PATTERN_COLOR_FORMAT), 1);
+ PUSH_DATA (push, NV50_2D_PATTERN_COLOR_FORMAT_32BPP);
+ BEGIN_NV04(push, NV50_2D(PATTERN_COLOR(0)), 4);
+ PUSH_DATA (push, 0x00000000);
+ PUSH_DATA (push, mask);
+ PUSH_DATA (push, 0xffffffff);
+ PUSH_DATA (push, 0xffffffff);
+ BEGIN_NV04(push, NV50_2D(OPERATION), 1);
+ PUSH_DATA (push, NV50_2D_OPERATION_ROP);
+ }
+
+ if (src->ms_x > dst->ms_x || src->ms_y > dst->ms_y) {
+ /* ms_x is always >= ms_y */
+ du_dx <<= src->ms_x - dst->ms_x;
+ dv_dy <<= src->ms_y - dst->ms_y;
+ } else {
+ du_dx >>= dst->ms_x - src->ms_x;
+ dv_dy >>= dst->ms_y - src->ms_y;
+ }
+
+ srcx = (int64_t)(info->src.box.x + srcx_adj) << (src->ms_x + 32);
+ srcy = (int64_t)(info->src.box.y + srcy_adj) << (src->ms_y + 32);
+
+ if (src->base.base.nr_samples > dst->base.base.nr_samples) {
+ /* center src coorinates for proper MS resolve filtering */
+ srcx += (int64_t)src->ms_x << 32;
+ srcy += (int64_t)src->ms_y << 32;
+ }
+
+ dstx = info->dst.box.x << dst->ms_x;
+ dsty = info->dst.box.y << dst->ms_y;
+
+ dstw = info->dst.box.width << dst->ms_x;
+ dsth = info->dst.box.height << dst->ms_y;
+
+ if (dstx < 0) {
+ dstw += dstx;
+ srcx -= du_dx * dstx;
+ dstx = 0;
+ }
+ if (dsty < 0) {
+ dsth += dsty;
+ srcy -= dv_dy * dsty;
+ dsty = 0;
+ }
+
+ BEGIN_NV04(push, NV50_2D(BLIT_CONTROL), 1);
+ PUSH_DATA (push, mode);
+ BEGIN_NV04(push, NV50_2D(BLIT_DST_X), 4);
+ PUSH_DATA (push, dstx);
+ PUSH_DATA (push, dsty);
+ PUSH_DATA (push, dstw);
+ PUSH_DATA (push, dsth);
+ BEGIN_NV04(push, NV50_2D(BLIT_DU_DX_FRACT), 4);
+ PUSH_DATA (push, du_dx);
+ PUSH_DATA (push, du_dx >> 32);
+ PUSH_DATA (push, dv_dy);
+ PUSH_DATA (push, dv_dy >> 32);
+
+ BCTX_REFN(nv50->bufctx, 2D, &dst->base, WR);
+ BCTX_REFN(nv50->bufctx, 2D, &src->base, RD);
+ nouveau_pushbuf_bufctx(nv50->base.pushbuf, nv50->bufctx);
+ if (nouveau_pushbuf_validate(nv50->base.pushbuf))
+ return;
+
+ for (i = 0; i < info->dst.box.depth; ++i) {
+ if (i > 0) {
+ /* no scaling in z-direction possible for eng2d blits */
+ if (dst->layout_3d) {
+ BEGIN_NV04(push, NV50_2D(DST_LAYER), 1);
+ PUSH_DATA (push, info->dst.box.z + i);
+ } else {
+ const unsigned z = info->dst.box.z + i;
+ BEGIN_NV04(push, NV50_2D(DST_ADDRESS_HIGH), 2);
+ PUSH_DATAh(push, dst->base.address + z * dst->layer_stride);
+ PUSH_DATA (push, dst->base.address + z * dst->layer_stride);
+ }
+ if (src->layout_3d) {
+ /* not possible because of depth tiling */
+ assert(0);
+ } else {
+ const unsigned z = info->src.box.z + i;
+ BEGIN_NV04(push, NV50_2D(SRC_ADDRESS_HIGH), 2);
+ PUSH_DATAh(push, src->base.address + z * src->layer_stride);
+ PUSH_DATA (push, src->base.address + z * src->layer_stride);
+ }
+ BEGIN_NV04(push, NV50_2D(BLIT_SRC_Y_INT), 1); /* trigger */
+ PUSH_DATA (push, srcy >> 32);
+ } else {
+ BEGIN_NV04(push, NV50_2D(BLIT_SRC_X_FRACT), 4);
+ PUSH_DATA (push, srcx);
+ PUSH_DATA (push, srcx >> 32);
+ PUSH_DATA (push, srcy);
+ PUSH_DATA (push, srcy >> 32);
+ }
+ }
+ nv50_bufctx_fence(nv50->bufctx, FALSE);
+
+ nouveau_bufctx_reset(nv50->bufctx, NV50_BIND_2D);
+
+ if (info->scissor_enable) {
+ BEGIN_NV04(push, NV50_2D(CLIP_ENABLE), 1);
+ PUSH_DATA (push, 0);
+ }
+ if (mask != 0xffffffff) {
+ BEGIN_NV04(push, NV50_2D(OPERATION), 1);
+ PUSH_DATA (push, NV50_2D_OPERATION_SRCCOPY);
+ }
+}
+
+static void
+nv50_blit(struct pipe_context *pipe, const struct pipe_blit_info *info)
+{
+ struct nv50_context *nv50 = nv50_context(pipe);
+ boolean eng3d = FALSE;
+
+ if (util_format_is_depth_or_stencil(info->dst.resource->format)) {
+ if (!(info->mask & PIPE_MASK_ZS))
+ return;
+ if (info->dst.resource->format == PIPE_FORMAT_Z32_FLOAT ||
+ info->dst.resource->format == PIPE_FORMAT_Z32_FLOAT_S8X24_UINT)
+ eng3d = TRUE;
+ if (info->filter != PIPE_TEX_FILTER_NEAREST)
+ eng3d = TRUE;
+ } else {
+ if (!(info->mask & PIPE_MASK_RGBA))
+ return;
+ if (info->mask != PIPE_MASK_RGBA)
+ eng3d = TRUE;
+ }
+
+ if (nv50_miptree(info->src.resource)->layout_3d) {
+ eng3d = TRUE;
+ } else
+ if (info->src.box.depth != info->dst.box.depth) {
+ eng3d = TRUE;
+ debug_printf("blit: cannot filter array or cube textures in z direction");
+ }
+
+ if (!eng3d && info->dst.format != info->src.format)
+ if (!nv50_2d_format_faithful(info->dst.format) ||
+ !nv50_2d_format_faithful(info->src.format))
+ eng3d = TRUE;
+
+ if (info->src.resource->nr_samples == 8 &&
+ info->dst.resource->nr_samples <= 1)
+ eng3d = TRUE;
+
+ /* FIXME: can't make this work with eng2d anymore */
+ if (info->src.resource->nr_samples > 1 ||
+ info->dst.resource->nr_samples > 1)
+ eng3d = TRUE;
+
+ /* FIXME: find correct src coordinate adjustments */
+ if ((info->src.box.width != info->dst.box.width &&
+ info->src.box.width != -info->dst.box.width) ||
+ (info->src.box.height != info->dst.box.height &&
+ info->src.box.height != -info->dst.box.height))
+ eng3d = TRUE;
+
+ if (!eng3d)
+ nv50_blit_eng2d(nv50, info);
+ else
+ nv50_blit_3d(nv50, info);
}
boolean
-nv50_blitctx_create(struct nv50_screen *screen)
+nv50_blitter_create(struct nv50_screen *screen)
{
- screen->blitctx = CALLOC_STRUCT(nv50_blitctx);
- if (!screen->blitctx) {
- NOUVEAU_ERR("failed to allocate blit context\n");
+ screen->blitter = CALLOC_STRUCT(nv50_blitter);
+ if (!screen->blitter) {
+ NOUVEAU_ERR("failed to allocate blitter struct\n");
return FALSE;
}
- screen->blitctx->screen = screen;
+ pipe_mutex_init(screen->blitter->mutex);
+
+ nv50_blitter_make_vp(screen->blitter);
+ nv50_blitter_make_sampler(screen->blitter);
+
+ return TRUE;
+}
+
+void
+nv50_blitter_destroy(struct nv50_screen *screen)
+{
+ struct nv50_blitter *blitter = screen->blitter;
+ unsigned i, m;
+
+ for (i = 0; i < NV50_BLIT_MAX_TEXTURE_TYPES; ++i) {
+ for (m = 0; m < NV50_BLIT_MODES; ++m) {
+ struct nv50_program *prog = blitter->fp[i][m];
+ if (prog) {
+ nv50_program_destroy(NULL, prog);
+ FREE((void *)prog->pipe.tokens);
+ FREE(prog);
+ }
+ }
+ }
- nv50_blitctx_make_vp(screen->blitctx);
- nv50_blitctx_make_fp(screen->blitctx);
+ FREE(blitter);
+}
- nv50_blitctx_make_sampler(screen->blitctx);
+boolean
+nv50_blitctx_create(struct nv50_context *nv50)
+{
+ nv50->blit = CALLOC_STRUCT(nv50_blitctx);
+ if (!nv50->blit) {
+ NOUVEAU_ERR("failed to allocate blit context\n");
+ return FALSE;
+ }
- screen->blitctx->color_mask = 0x1111;
+ nv50->blit->nv50 = nv50;
return TRUE;
}
struct pipe_context *pipe = &nv50->base.pipe;
pipe->resource_copy_region = nv50_resource_copy_region;
- pipe->resource_resolve = nv50_resource_resolve;
+ pipe->blit = nv50_blit;
pipe->clear_render_target = nv50_clear_render_target;
pipe->clear_depth_stencil = nv50_clear_depth_stencil;
}
projects / mesa.git / blobdiff