nvfx: refactor sampling code, add support for swizzles and depth tex

This is a significant refactoring of the sampling code that:
- Moves all generic functions in nvfx_fragtex.c
- Adds a driver-specific sampler view structure and uses it to
  precompute texture setup as it should be done
- Unifies a bit more of code between nv30 and nv40
- Adds support for sampler view swizzles
- Support for specifying as sampler view format different from the
  resource one (only trivially)
- Support for sampler view specification of first and last level
- Support for depth textures on nv30, both for reading depth and
  for compare
- Support for sRGB textures
- Unifies the format table between nv30 and nv40
- Expands the format table to include essentially all supportable formats
  except mixed sign and "autonormal" formats
- Fixes the "is format supported" logic, which was quite broken, and
  makes it use the format table

Only tested on nv30 currently.

1 files changed, 66 insertions, 108 deletions

diff --git a/src/gallium/drivers/nvfx/nv30_fragtex.c b/src/gallium/drivers/nvfx/nv30_fragtex.c
index fbc69cf44a..63c578a0ce 100644
--- a/src/gallium/drivers/nvfx/nv30_fragtex.c
+++ b/src/gallium/drivers/nvfx/nv30_fragtex.c
@@ -10,88 +10,75 @@ nv30_sampler_state_init(struct pipe_context *pipe,
 			  struct nvfx_sampler_state *ps,
 			  const struct pipe_sampler_state *cso)
 {
-	if (cso->max_anisotropy >= 8) {
-		ps->en |= NV34TCL_TX_ENABLE_ANISO_8X;
-	} else
-	if (cso->max_anisotropy >= 4) {
-		ps->en |= NV34TCL_TX_ENABLE_ANISO_4X;
-	} else
-	if (cso->max_anisotropy >= 2) {
-		ps->en |= NV34TCL_TX_ENABLE_ANISO_2X;
-	}
+	float limit;
 
+	if (cso->max_anisotropy >= 2)
 	{
-		float limit;
+		if (cso->max_anisotropy >= 8)
+			ps->en |= NV34TCL_TX_ENABLE_ANISO_8X;
+		else if (cso->max_anisotropy >= 4)
+			ps->en |= NV34TCL_TX_ENABLE_ANISO_4X;
+		else if (cso->max_anisotropy >= 2)
+			ps->en |= NV34TCL_TX_ENABLE_ANISO_2X;
+	}
 
-		limit = CLAMP(cso->lod_bias, -16.0, 15.0);
-		ps->filt |= (int)(cso->lod_bias * 256.0) & 0x1fff;
+	limit = CLAMP(cso->lod_bias, -16.0, 15.0);
+	ps->filt |= (int)(cso->lod_bias * 256.0) & 0x1fff;
 
-		limit = CLAMP(cso->max_lod, 0.0, 15.0);
-		ps->en |= (int)(limit) << 14 /*NV34TCL_TX_ENABLE_MIPMAP_MAX_LOD_SHIFT*/;
+	ps->max_lod = (int)CLAMP(cso->max_lod, 0.0, 15.0);
+	ps->min_lod = (int)CLAMP(cso->min_lod, 0.0, 15.0);
 
-		limit = CLAMP(cso->min_lod, 0.0, 15.0);
-		ps->en |= (int)(limit) << 26 /*NV34TCL_TX_ENABLE_MIPMAP_MIN_LOD_SHIFT*/;
-	}
+	ps->en |= NV34TCL_TX_ENABLE_ENABLE;
 }
 
-#define _(m,tf,ts0x,ts0y,ts0z,ts0w,ts1x,ts1y,ts1z,ts1w)                        \
-[PIPE_FORMAT_##m] = {                                                                              \
-  NV34TCL_TX_FORMAT_FORMAT_##tf,                                               \
-  NV34TCL_TX_FORMAT_FORMAT_##tf##_RECT,                                               \
-  (NV34TCL_TX_SWIZZLE_S0_X_##ts0x | NV34TCL_TX_SWIZZLE_S0_Y_##ts0y |           \
-   NV34TCL_TX_SWIZZLE_S0_Z_##ts0z | NV34TCL_TX_SWIZZLE_S0_W_##ts0w |           \
-   NV34TCL_TX_SWIZZLE_S1_X_##ts1x | NV34TCL_TX_SWIZZLE_S1_Y_##ts1y |           \
-   NV34TCL_TX_SWIZZLE_S1_Z_##ts1z | NV34TCL_TX_SWIZZLE_S1_W_##ts1w)            \
-}
+void
+nv30_sampler_view_init(struct pipe_context *pipe,
+			  struct nvfx_sampler_view *sv)
+{
+	struct pipe_resource* pt = sv->base.texture;
+	struct nvfx_texture_format *tf = &nvfx_texture_formats[sv->base.format];
+	unsigned txf;
+	unsigned level = pt->target == PIPE_TEXTURE_CUBE ? 0 : sv->base.first_level;
+
+	assert(tf->fmt[0] >= 0);
+
+	txf = sv->u.init_fmt;
+	txf |= (level != sv->base.last_level ? NV34TCL_TX_FORMAT_MIPMAP : 0);
+	txf |= log2i(u_minify(pt->width0, level)) << NV34TCL_TX_FORMAT_BASE_SIZE_U_SHIFT;
+	txf |= log2i(u_minify(pt->height0, level)) << NV34TCL_TX_FORMAT_BASE_SIZE_V_SHIFT;
+	txf |= log2i(u_minify(pt->depth0, level)) << NV34TCL_TX_FORMAT_BASE_SIZE_W_SHIFT;
+	txf |=  0x10000;
+
+	sv->u.nv30.fmt[0] = tf->fmt[0] | txf;
+	sv->u.nv30.fmt[1] = tf->fmt[1] | txf;
+	sv->u.nv30.fmt[2] = tf->fmt[2] | txf;
+	sv->u.nv30.fmt[3] = tf->fmt[3] | txf;
 
-struct nv30_texture_format {
-	int     format;
-	int     rect_format;
-	int     swizzle;
-};
-
-#define NV34TCL_TX_FORMAT_FORMAT_DXT1_RECT NV34TCL_TX_FORMAT_FORMAT_DXT1
-#define NV34TCL_TX_FORMAT_FORMAT_DXT3_RECT NV34TCL_TX_FORMAT_FORMAT_DXT3
-#define NV34TCL_TX_FORMAT_FORMAT_DXT5_RECT NV34TCL_TX_FORMAT_FORMAT_DXT5
-
-static struct nv30_texture_format
-nv30_texture_formats[PIPE_FORMAT_COUNT] = {
-	[0 ... PIPE_FORMAT_COUNT - 1] = {-1, 0, 0},
-	_(B8G8R8X8_UNORM, A8R8G8B8,   S1,   S1,   S1,  ONE, X, Y, Z, W),
-	_(B8G8R8A8_UNORM, A8R8G8B8,   S1,   S1,   S1,   S1, X, Y, Z, W),
-	_(B5G5R5A1_UNORM, A1R5G5B5,   S1,   S1,   S1,   S1, X, Y, Z, W),
-	_(B4G4R4A4_UNORM, A4R4G4B4,   S1,   S1,   S1,   S1, X, Y, Z, W),
-	_(B5G6R5_UNORM  , R5G6B5  ,   S1,   S1,   S1,  ONE, X, Y, Z, W),
-	_(L8_UNORM      , L8      ,   S1,   S1,   S1,  ONE, X, X, X, X),
-	_(A8_UNORM      , L8      , ZERO, ZERO, ZERO,   S1, X, X, X, X),
-	_(I8_UNORM      , L8      ,   S1,   S1,   S1,   S1, X, X, X, X),
-	_(L8A8_UNORM    , A8L8    ,   S1,   S1,   S1,   S1, X, X, X, Y),
-	_(Z16_UNORM     , R5G6B5  ,   S1,   S1,   S1,  ONE, X, X, X, X),
-	_(S8_USCALED_Z24_UNORM   , A8R8G8B8,   S1,   S1,   S1,  ONE, X, X, X, X),
-	_(DXT1_RGB      , DXT1    ,   S1,   S1,   S1,  ONE, X, Y, Z, W),
-	_(DXT1_RGBA     , DXT1    ,   S1,   S1,   S1,   S1, X, Y, Z, W),
-	_(DXT3_RGBA     , DXT3    ,   S1,   S1,   S1,   S1, X, Y, Z, W),
-	_(DXT5_RGBA     , DXT5    ,   S1,   S1,   S1,   S1, X, Y, Z, W),
-	{},
-};
+	sv->swizzle  |= (nvfx_subresource_pitch(pt, 0) << NV34TCL_TX_SWIZZLE_RECT_PITCH_SHIFT);
+
+	if(pt->height0 <= 1 || util_format_is_compressed(sv->base.format))
+		sv->u.nv30.rect = -1;
+	else
+		sv->u.nv30.rect = !!(pt->flags & NVFX_RESOURCE_FLAG_LINEAR);
+
+	sv->lod_offset = sv->base.first_level - level;
+	sv->max_lod_limit = sv->base.last_level - level;
+}
 
 void
 nv30_fragtex_set(struct nvfx_context *nvfx, int unit)
 {
 	struct nvfx_sampler_state *ps = nvfx->tex_sampler[unit];
-	struct nvfx_miptree *mt = (struct nvfx_miptree *)nvfx->fragment_sampler_views[unit]->texture;
-	struct pipe_resource *pt = &mt->base.base;
-	struct nouveau_bo *bo = mt->base.bo;
-	struct nv30_texture_format *tf;
+	struct nvfx_sampler_view* sv = (struct nvfx_sampler_view*)nvfx->fragment_sampler_views[unit];
+	struct nouveau_bo *bo = ((struct nvfx_miptree *)sv->base.texture)->base.bo;
 	struct nouveau_channel* chan = nvfx->screen->base.channel;
-	uint32_t txf, txs;
+	unsigned txf;
 	unsigned tex_flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD;
 	unsigned use_rect;
+	unsigned max_lod = MIN2(ps->max_lod + sv->lod_offset, sv->max_lod_limit);
+	unsigned min_lod = MIN2(ps->min_lod + sv->lod_offset, max_lod) ;
 
-	tf = &nv30_texture_formats[pt->format];
-	assert(tf->format >= 0);
-
-	if(pt->height0 <= 1 || util_format_is_compressed(pt->format))
+	if(sv->u.nv30.rect < 0)
 	{
 		/* in the case of compressed or 1D textures, we can get away with this,
 		 * since the layout is the same
@@ -100,9 +87,9 @@ nv30_fragtex_set(struct nvfx_context *nvfx, int unit)
 	}
 	else
 	{
-		static int warned = 0;
-		if(!warned && !ps->fmt != !(pt->flags & NVFX_RESOURCE_FLAG_LINEAR)) {
-			warned = 1;
+		static boolean warned = FALSE;
+		if( !!ps->fmt != sv->u.nv30.rect && !warned) {
+			warned = TRUE;
 			fprintf(stderr,
 					"Unimplemented: coordinate normalization mismatch. Possible reasons:\n"
 					"1. ARB_texture_non_power_of_two is being used despite the fact it isn't supported\n"
@@ -110,51 +97,22 @@ nv30_fragtex_set(struct nvfx_context *nvfx, int unit)
 					"3. The state tracker is not supported\n");
 		}
 
-		use_rect  = pt->flags & NVFX_RESOURCE_FLAG_LINEAR;
-	}
-
-	txf = use_rect ? tf->rect_format : tf->format;
-	txf |= ((pt->last_level>0) ? NV34TCL_TX_FORMAT_MIPMAP : 0);
-	txf |= log2i(pt->width0) << NV34TCL_TX_FORMAT_BASE_SIZE_U_SHIFT;
-	txf |= log2i(pt->height0) << NV34TCL_TX_FORMAT_BASE_SIZE_V_SHIFT;
-	txf |= log2i(pt->depth0) << NV34TCL_TX_FORMAT_BASE_SIZE_W_SHIFT;
-	txf |= NV34TCL_TX_FORMAT_NO_BORDER | 0x10000;
-
-	switch (pt->target) {
-	case PIPE_TEXTURE_CUBE:
-		txf |= NV34TCL_TX_FORMAT_CUBIC;
-		/* fall-through */
-	case PIPE_TEXTURE_2D:
-	case PIPE_TEXTURE_RECT:
-		txf |= NV34TCL_TX_FORMAT_DIMS_2D;
-		break;
-	case PIPE_TEXTURE_3D:
-		txf |= NV34TCL_TX_FORMAT_DIMS_3D;
-		break;
-	case PIPE_TEXTURE_1D:
-		txf |= NV34TCL_TX_FORMAT_DIMS_1D;
-		break;
-	default:
-		NOUVEAU_ERR("Unknown target %d\n", pt->target);
-		return;
+		use_rect  = sv->u.nv30.rect;
 	}
 
-	txs = tf->swizzle;
-
-	if(use_rect)
-		txs |= nvfx_subresource_pitch(&mt->base, 0) << NV34TCL_TX_SWIZZLE_RECT_PITCH_SHIFT;
+	txf = sv->u.nv30.fmt[ps->compare + (use_rect ? 2 : 0)];
 
 	MARK_RING(chan, 9, 2);
 	OUT_RING(chan, RING_3D(NV34TCL_TX_OFFSET(unit), 8));
-	OUT_RELOC(chan, bo, 0, tex_flags | NOUVEAU_BO_LOW, 0, 0);
-	OUT_RELOC(chan, bo, txf, tex_flags | NOUVEAU_BO_OR,
-		      NV34TCL_TX_FORMAT_DMA0, NV34TCL_TX_FORMAT_DMA1);
-	OUT_RING(chan, ps->wrap);
-	OUT_RING(chan, NV34TCL_TX_ENABLE_ENABLE | ps->en);
-	OUT_RING(chan, txs);
-	OUT_RING(chan, ps->filt | 0x2000 /*voodoo*/);
-	OUT_RING(chan, (pt->width0 << NV34TCL_TX_NPOT_SIZE_W_SHIFT) |
-		       pt->height0);
+	OUT_RELOC(chan, bo, sv->offset, tex_flags | NOUVEAU_BO_LOW, 0, 0);
+	OUT_RELOC(chan, bo, txf,
+		tex_flags | NOUVEAU_BO_OR,
+		NV34TCL_TX_FORMAT_DMA0, NV34TCL_TX_FORMAT_DMA1);
+	OUT_RING(chan, (ps->wrap & sv->wrap_mask) | sv->wrap);
+	OUT_RING(chan, ps->en | (min_lod << NV34TCL_TX_ENABLE_MIPMAP_MIN_LOD_SHIFT) | (max_lod << NV34TCL_TX_ENABLE_MIPMAP_MAX_LOD_SHIFT));
+	OUT_RING(chan, sv->swizzle);
+	OUT_RING(chan, ps->filt | sv->filt);
+	OUT_RING(chan, sv->npot_size);
 	OUT_RING(chan, ps->bcol);
 
 	nvfx->hw_txf[unit] = txf;