From: Rob Clark Date: Thu, 15 Dec 2016 22:27:56 +0000 (-0500) Subject: freedreno/a5xx: border color support X-Git-Url: https://git.libre-soc.org/?a=commitdiff_plain;h=2c0dfd48f0f510c757bc7b332d66cfed79125c2b;p=mesa.git freedreno/a5xx: border color support Not 100% sure it works if you have border color in VS.. but it might be right. Signed-off-by: Rob Clark --- diff --git a/src/gallium/drivers/freedreno/a5xx/fd5_emit.c b/src/gallium/drivers/freedreno/a5xx/fd5_emit.c index 35ecb61d24d..1b9336165dd 100644 --- a/src/gallium/drivers/freedreno/a5xx/fd5_emit.c +++ b/src/gallium/drivers/freedreno/a5xx/fd5_emit.c @@ -122,10 +122,158 @@ fd5_emit_const_bo(struct fd_ringbuffer *ring, enum shader_t type, boolean write, } } +/* Border color layout is diff from a4xx/a5xx.. if it turns out to be + * the same as a6xx then move this somewhere common ;-) + * + * Entry layout looks like (total size, 0x60 bytes): + * + * offset | description + * -------+------------- + * 0x00 | fp32[0] + * | fp32[1] + * | fp32[2] + * | fp32[3] + * 0x10 | uint16[0] + * | uint16[1] + * | uint16[2] + * | uint16[3] + * 0x18 | int16[0] + * | int16[1] + * | int16[2] + * | int16[3] + * 0x20 | fp16[0] + * | fp16[1] + * | fp16[2] + * | fp16[3] + * 0x28 | ?? maybe padding ?? + * 0x30 | uint8[0] + * | uint8[1] + * | uint8[2] + * | uint8[3] + * 0x34 | int8[0] + * | int8[1] + * | int8[2] + * | int8[3] + * 0x38 | ?? maybe padding ?? + * + * Some uncertainty, because not clear that this actually works properly + * with blob, so who knows.. + */ + +struct PACKED bcolor_entry { + uint32_t fp32[4]; + uint16_t ui16[4]; + int16_t si16[4]; + uint16_t fp16[4]; + uint8_t __pad0[8]; + uint8_t ui8[4]; + int8_t si8[4]; + uint8_t __pad1[40]; +}; + +#define FD5_BORDER_COLOR_SIZE 0x60 +#define FD5_BORDER_COLOR_UPLOAD_SIZE (2 * PIPE_MAX_SAMPLERS * FD5_BORDER_COLOR_SIZE) +#define FD5_BORDER_COLOR_OFFSET 8 /* TODO probably should be dynamic */ + +static void +setup_border_colors(struct fd_texture_stateobj *tex, struct bcolor_entry *entries) +{ + unsigned i, j; + + debug_assert(tex->num_samplers < FD5_BORDER_COLOR_OFFSET); // TODO + + for (i = 0; i < tex->num_samplers; i++) { + struct bcolor_entry *e = &entries[i]; + struct pipe_sampler_state *sampler = tex->samplers[i]; + union pipe_color_union *bc; + + if (!sampler) + continue; + + bc = &sampler->border_color; + + /* + * XXX HACK ALERT XXX + * + * The border colors need to be swizzled in a particular + * format-dependent order. Even though samplers don't know about + * formats, we can assume that with a GL state tracker, there's a + * 1:1 correspondence between sampler and texture. Take advantage + * of that knowledge. + */ + if ((i >= tex->num_textures) || !tex->textures[i]) + continue; + + const struct util_format_description *desc = + util_format_description(tex->textures[i]->format); + + for (j = 0; j < 4; j++) { + int c = desc->swizzle[j]; + + if (c >= 4) + continue; + + if (desc->channel[c].pure_integer) { + float f = bc->i[c]; + + e->fp32[j] = fui(f); + e->fp16[j] = util_float_to_half(f); + e->ui16[j] = bc->ui[c]; + e->si16[j] = bc->i[c]; + e->ui8[j] = bc->ui[c]; + e->si8[j] = bc->i[c]; + } else { + float f = bc->f[c]; + + e->fp32[j] = fui(f); + e->fp16[j] = util_float_to_half(f); + e->ui16[j] = f * 65535.0; + e->si16[j] = f * 32767.5; + e->ui8[j] = f * 255.0; + e->si8[j] = f * 128.0; + } + } + +#ifdef DEBUG + memset(&e->__pad0, 0, sizeof(e->__pad0)); + memset(&e->__pad1, 0, sizeof(e->__pad1)); +#endif + } +} + static void +emit_border_color(struct fd_context *ctx, struct fd_ringbuffer *ring) +{ + struct fd5_context *fd5_ctx = fd5_context(ctx); + struct bcolor_entry *entries; + unsigned off; + void *ptr; + + STATIC_ASSERT(sizeof(struct bcolor_entry) == FD5_BORDER_COLOR_SIZE); + + u_upload_alloc(fd5_ctx->border_color_uploader, + 0, FD5_BORDER_COLOR_UPLOAD_SIZE, + FD5_BORDER_COLOR_UPLOAD_SIZE, &off, + &fd5_ctx->border_color_buf, + &ptr); + + entries = ptr; + + setup_border_colors(&ctx->verttex, &entries[0]); + setup_border_colors(&ctx->fragtex, &entries[ctx->verttex.num_samplers]); + + OUT_PKT4(ring, REG_A5XX_TPL1_TP_BORDER_COLOR_BASE_ADDR_LO, 2); + OUT_RELOC(ring, fd_resource(fd5_ctx->border_color_buf)->bo, off, 0, 0); + + u_upload_unmap(fd5_ctx->border_color_uploader); +} + +static bool emit_textures(struct fd_context *ctx, struct fd_ringbuffer *ring, enum adreno_state_block sb, struct fd_texture_stateobj *tex) { + bool needs_border = false; + unsigned bcolor_offset = (sb == SB_FRAG_TEX) ? ctx->verttex.num_samplers : 0; unsigned i; if (tex->num_samplers > 0) { @@ -145,8 +293,11 @@ emit_textures(struct fd_context *ctx, struct fd_ringbuffer *ring, &dummy_sampler; OUT_RING(ring, sampler->texsamp0); OUT_RING(ring, sampler->texsamp1); - OUT_RING(ring, sampler->texsamp2); + OUT_RING(ring, sampler->texsamp2 | + A5XX_TEX_SAMP_2_BCOLOR_OFFSET(bcolor_offset)); OUT_RING(ring, sampler->texsamp3); + + needs_border |= sampler->needs_border; } } @@ -188,6 +339,8 @@ emit_textures(struct fd_context *ctx, struct fd_ringbuffer *ring, OUT_RING(ring, view->texconst11); } } + + return needs_border; } void @@ -241,6 +394,7 @@ fd5_emit_state(struct fd_context *ctx, struct fd_ringbuffer *ring, const struct ir3_shader_variant *vp = fd5_emit_get_vp(emit); const struct ir3_shader_variant *fp = fd5_emit_get_fp(emit); uint32_t dirty = emit->dirty; + bool needs_border = false; emit_marker5(ring, 5); @@ -444,7 +598,7 @@ fd5_emit_state(struct fd_context *ctx, struct fd_ringbuffer *ring, if (dirty & FD_DIRTY_VERTTEX) { if (vp->has_samp) { - emit_textures(ctx, ring, SB_VERT_TEX, &ctx->verttex); + needs_border |= emit_textures(ctx, ring, SB_VERT_TEX, &ctx->verttex); OUT_PKT4(ring, REG_A5XX_TPL1_VS_TEX_COUNT, 1); OUT_RING(ring, ctx->verttex.num_textures); } else { @@ -454,7 +608,7 @@ fd5_emit_state(struct fd_context *ctx, struct fd_ringbuffer *ring, if (dirty & FD_DIRTY_FRAGTEX) { if (fp->has_samp) { - emit_textures(ctx, ring, SB_FRAG_TEX, &ctx->fragtex); + needs_border |= emit_textures(ctx, ring, SB_FRAG_TEX, &ctx->fragtex); OUT_PKT4(ring, REG_A5XX_TPL1_FS_TEX_COUNT, 1); OUT_RING(ring, ctx->fragtex.num_textures); } else { @@ -462,6 +616,9 @@ fd5_emit_state(struct fd_context *ctx, struct fd_ringbuffer *ring, } } + if (needs_border) + emit_border_color(ctx, ring); + ctx->dirty &= ~dirty; }