* Kill the A8+A8 hack. Recent enough X servers (>=1.7) fall back to
ARGB glyphs for drivers not supporting A8 render targets.
* Kill all the global state. It doesn't matter a lot yet but it might
if we get multicard working at some point.
* Other random clean-ups with no functional changes.
Some numbers from x11perf -aa10text -aa24text -comppixwin10 -comppixwin500:
* Before, with A glyphs:
12000000 trep @ 0.0025 msec (394000.0/sec): Char in 80-char aa line
(Charter 10)
3200000 trep @ 0.0119 msec ( 84300.0/sec): Char in 30-char aa line
(Charter 24)
8000000 trep @ 0.0036 msec (280000.0/sec): Composite 10x10 from pixmap to
window
60000 trep @ 0.4813 msec ( 2080.0/sec): Composite 500x500 from pixmap
to window
* After, ARGB glyphs:
20000000 trep @ 0.0015 msec (649000.0/sec): Char in 80-char aa line
(Charter 10)
4800000 trep @ 0.0057 msec (175000.0/sec): Char in 30-char aa line
(Charter 24)
8000000 trep @ 0.0036 msec (280000.0/sec): Composite 10x10 from pixmap to
window
60000 trep @ 0.4813 msec ( 2080.0/sec): Composite 500x500 from pixmap
to window
Signed-off-by: Francisco Jerez <currojerez at riseup.net>
---
src/nv10_exa.c | 890 ++++++++++++++++++++++----------------------------------
1 files changed, 345 insertions(+), 545 deletions(-)
diff --git a/src/nv10_exa.c b/src/nv10_exa.c
index 3665230..6216730 100644
--- a/src/nv10_exa.c
+++ b/src/nv10_exa.c
@@ -2,6 +2,7 @@
* Copyright 2007 Stephane Marchesin
* Copyright 2007 Arthur Huillet
* Copyright 2007 Peter Winters
+ * Copyright 2009 Francisco Jerez
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the
"Software"),
@@ -28,24 +29,83 @@
#include "nv_include.h"
-typedef struct nv10_exa_state {
- Bool have_mask;
- Bool is_a8_plus_a8;
- struct {
- PictTransformPtr transform;
- float width;
- float height;
- } unit[2];
-} nv10_exa_state_t;
-static nv10_exa_state_t state;
+/* Texture/Render target formats. */
+static struct pict_format {
+ int exa;
+ int hw;
+} nv10_tex_format_pot[] = {
+ { PICT_a8, 0x80 },
+ { PICT_r5g6b5, 0x280 },
+ { PICT_x8r8g8b8, 0x300 },
+ { PICT_a8r8g8b8, 0x300 },
+ {},
+
+}, nv10_tex_format_rect[] = {
+ { PICT_a8, 0x980 },
+ { PICT_r5g6b5, 0x880 },
+ { PICT_x8r8g8b8, 0x900 },
+ { PICT_a8r8g8b8, 0x900 },
+ {},
+
+}, nv20_tex_format_rect[] = {
+ { PICT_a8, 0xd80 },
+ { PICT_r5g6b5, 0x880 },
+ { PICT_x8r8g8b8, 0x900 },
+ { PICT_a8r8g8b8, 0x900 },
+ {},
+
+}, nv10_rt_format[] = {
+ { PICT_r5g6b5, 0x103 },
+ { PICT_x8r8g8b8, 0x108 },
+ { PICT_a8r8g8b8, 0x108 },
+ {},
+};
+
+static int
+get_tex_format(PicturePtr pict)
+{
+ ScrnInfoPtr pScrn = xf86Screens[pict->pDrawable->pScreen->myNum];
+ NVPtr pNv = NVPTR(pScrn);
+
+ /* If repeat is set we're always handling a 1x1 texture with
+ * ARGB/XRGB destination, in that case we change the format to
+ * use the POT (swizzled) matching format.
+ */
+ struct pict_format *format + pict->repeat != RepeatNone ?
nv10_tex_format_pot :
+ pNv->Architecture == NV_ARCH_20 ? nv20_tex_format_rect :
+ nv10_tex_format_rect;
+
+ for (; format->hw; format++) {
+ if (format->exa == pict->format)
+ return format->hw;
+ }
+
+ return 0;
+}
+static int
+get_rt_format(PicturePtr pict)
+{
+ struct pict_format *format = nv10_rt_format;
+
+ for (; format->hw; format++) {
+ if (format->exa == pict->format)
+ return format->hw;
+ }
+
+ return 0;
+}
+
+/* Blending functions. */
#define SF(x) NV10TCL_BLEND_FUNC_SRC_##x
#define DF(x) NV10TCL_BLEND_FUNC_DST_##x
-static struct nv10_pictop {
+static struct pict_op {
int src;
int dst;
-} NV10PictOp [] = {
+
+} nv10_pict_op[] = {
{ SF(ZERO), DF(ZERO) }, /* Clear */
{ SF(ONE), DF(ZERO) }, /* Src */
{ SF(ZERO), DF(ONE) }, /* Dst */
@@ -61,131 +121,91 @@ static struct nv10_pictop {
{ SF(ONE), DF(ONE) }, /* Add */
};
-static inline bool needs_src_alpha(int op)
+static inline Bool
+needs_src_alpha(int op)
{
- return NV10PictOp[op].dst == DF(ONE_MINUS_SRC_ALPHA)
- || NV10PictOp[op].dst == DF(SRC_ALPHA);
+ return nv10_pict_op[op].dst == DF(ONE_MINUS_SRC_ALPHA)
+ || nv10_pict_op[op].dst == DF(SRC_ALPHA);
}
-static inline bool needs_src(int op)
+static inline Bool
+needs_src(int op)
{
- return NV10PictOp[op].src != DF(ZERO);
+ return nv10_pict_op[op].src != DF(ZERO);
}
-static inline bool effective_component_alpha(PicturePtr mask)
+static inline Bool
+effective_component_alpha(PicturePtr mask)
{
return mask && mask->componentAlpha &&
PICT_FORMAT_RGB(mask->format);
}
-static int NV10TexFormat(int ExaFormat)
-{
- struct {int exa;int hw;} tex_format[] - {
- {PICT_a8r8g8b8, 0x900},
- {PICT_x8r8g8b8, 0x900},
- {PICT_r5g6b5, 0x880}, /*this one was only tested with rendercheck*/
- /*{PICT_a1r5g5b5, NV10TCL_TX_FORMAT_FORMAT_R5G5B5A1},
- {PICT_a4r4g4b4, NV10TCL_TX_FORMAT_FORMAT_R4G4B4A4},*/
- {PICT_a8, 0x980}, /*this is a NV1x only format, corresponding NV2x is 0xD80,
we hack it in below*/
- };
-
- int i;
- for (i = 0; i < sizeof(tex_format) / sizeof(tex_format[0]); i++) {
- if (tex_format[i].exa == ExaFormat)
- return tex_format[i].hw;
- }
-
- return 0;
-}
-
-static int NV10DstFormat(int ExaFormat)
-{
- struct {int exa;int hw;} dst_format[] - {
- {PICT_a8r8g8b8, 0x108},
- {PICT_x8r8g8b8, 0x108},
- {PICT_r5g6b5, 0x103}
- };
-
- int i;
- for (i = 0; i < sizeof(dst_format) / sizeof(dst_format[0]); i++) {
- if (dst_format[i].exa == ExaFormat)
- return dst_format[i].hw;
- }
-
- return 0;
-}
-
-static Bool NV10CheckTexture(PicturePtr Picture)
+static Bool
+check_texture(PicturePtr pict)
{
int w, h;
- if (!Picture->pDrawable)
+ if (!pict->pDrawable)
NOUVEAU_FALLBACK("Solid and gradient pictures unsupported\n");
- w = Picture->pDrawable->width;
- h = Picture->pDrawable->height;
+ w = pict->pDrawable->width;
+ h = pict->pDrawable->height;
- if ((w > 2046) || (h > 2046))
+ if (w > 2046 || h > 2046)
NOUVEAU_FALLBACK("picture too large, %dx%d\n", w, h);
- if (!NV10TexFormat(Picture->format))
+ if (!get_tex_format(pict))
return FALSE;
- if (Picture->filter != PictFilterNearest && Picture->filter !=
PictFilterBilinear)
+
+ if (pict->filter != PictFilterNearest &&
+ pict->filter != PictFilterBilinear)
return FALSE;
- /* we cannot repeat on NV10 because NPOT textures do not support this.
unfortunately. */
- if (Picture->repeat != RepeatNone)
+
+ /* We cannot repeat on NV10 because NPOT textures do not
+ * support this. unfortunately. */
+ if (pict->repeat != RepeatNone)
/* we can repeat 1x1 textures */
if (!(w == 1 && h == 1))
return FALSE;
+
return TRUE;
}
-static Bool NV10CheckBuffer(PicturePtr Picture)
+static Bool
+check_render_target(PicturePtr pict)
{
- int w = Picture->pDrawable->width;
- int h = Picture->pDrawable->height;
+ int w = pict->pDrawable->width;
+ int h = pict->pDrawable->height;
- if ((w > 4096) || (h > 4096))
+ if (w > 4096 || h > 4096)
return FALSE;
- if (!NV10DstFormat(Picture->format))
- return FALSE;
- return TRUE;
-}
-static Bool NV10CheckPictOp(int op)
-{
- if (op >= PictOpSaturate) {
- /*we do no saturate, disjoint, conjoint, though we could do e.g.
DisjointClear which really is Clear */
+ if (!get_rt_format(pict))
return FALSE;
- }
+
return TRUE;
}
-/* Check if the current operation is a doable A8 + A8 */
-/* A8 destination is a special case, because we do it by having the card think
-it's ARGB. For now we support PictOpAdd which is the only important op for
this dst format,
-and without transformation or funny things.*/
-static Bool NV10Check_A8plusA8_Feasability(PicturePtr src, PicturePtr msk,
PicturePtr dst, int op)
+static Bool
+check_pict_op(int op)
{
-#if X_BYTE_ORDER == X_BIG_ENDIAN
- return FALSE;
-#endif
- if ((!msk) && (src->format == PICT_a8) && (dst->format
== PICT_a8) && (!src->transform) &&
- (op == PictOpAdd) && (src->repeat == RepeatNone)) {
- return TRUE;
- }
- return FALSE;
+ /* We do no saturate, disjoint, conjoint, though we
+ * could do e.g. DisjointClear which really is
+ * Clear. */
+ return op < PictOpSaturate;
}
#if 0
-static void NV10EXAFallbackInfo(char *reason, int op, PicturePtr pSrcPicture,
- PicturePtr pMaskPicture, PicturePtr pDstPicture)
+static void
+print_fallback_info(char *reason, int op, PicturePtr src, PicturePtr mask,
+ PicturePtr dst)
{
char out2[4096];
char *out = out2;
+
sprintf(out, "%s ", reason);
- out = out + strlen(out);
+ out += strlen(out);
+
switch (op) {
case PictOpClear:
sprintf(out, "PictOpClear ");
@@ -208,8 +228,9 @@ static void NV10EXAFallbackInfo(char *reason, int op,
PicturePtr pSrcPicture,
default:
sprintf(out, "PictOp%d ", op);
}
- out = out + strlen(out);
- switch (pSrcPicture->format) {
+ out += strlen(out);
+
+ switch (src->format) {
case PICT_a8r8g8b8:
sprintf(out, "A8R8G8B8 ");
break;
@@ -229,16 +250,18 @@ static void NV10EXAFallbackInfo(char *reason, int op,
PicturePtr pSrcPicture,
sprintf(out, "A1 ");
break;
default:
- sprintf(out, "%x ", pSrcPicture->format);
+ sprintf(out, "%x ", src->format);
}
out += strlen(out);
- sprintf(out, "(%dx%d) ", pSrcPicture->pDrawable->width,
pSrcPicture->pDrawable->height);
- if (pSrcPicture->repeat != RepeatNone)
+
+ sprintf(out, "(%dx%d) ", src->pDrawable->width,
+ src->pDrawable->height);
+ if (src->repeat != RepeatNone)
strcat(out, "R ");
strcat(out, "-> ");
out += strlen(out);
- switch (pDstPicture->format) {
+ switch (dst->format) {
case PICT_a8r8g8b8:
sprintf(out, "A8R8G8B8 ");
break;
@@ -258,17 +281,20 @@ static void NV10EXAFallbackInfo(char *reason, int op,
PicturePtr pSrcPicture,
sprintf(out, "A1 ");
break;
default:
- sprintf(out, "%x ", pDstPicture->format);
+ sprintf(out, "%x ", dst->format);
}
out += strlen(out);
- sprintf(out, "(%dx%d) ", pDstPicture->pDrawable->width,
pDstPicture->pDrawable->height);
- if (pDstPicture->repeat != RepeatNone)
+
+ sprintf(out, "(%dx%d) ", dst->pDrawable->width,
+ dst->pDrawable->height);
+ if (dst->repeat != RepeatNone)
strcat(out, "R ");
out += strlen(out);
- if (!pMaskPicture)
+
+ if (!mask)
sprintf(out, "& NONE");
else {
- switch (pMaskPicture->format) {
+ switch (mask->format) {
case PICT_a8r8g8b8:
sprintf(out, "& A8R8G8B8 ");
break;
@@ -285,115 +311,97 @@ static void NV10EXAFallbackInfo(char *reason, int op,
PicturePtr pSrcPicture,
sprintf(out, "& A1 ");
break;
default:
- sprintf(out, "& %x ", pMaskPicture->format);
+ sprintf(out, "& %x ", mask->format);
}
out += strlen(out);
- sprintf(out, "(%dx%d) ", pMaskPicture->pDrawable->width,
pMaskPicture->pDrawable->height);
- if (pMaskPicture->repeat != RepeatNone)
+
+ sprintf(out, "(%dx%d) ", mask->pDrawable->width,
+ mask->pDrawable->height);
+ if (mask->repeat != RepeatNone)
strcat(out, "R ");
- if (pMaskPicture->componentAlpha)
+ if (mask->componentAlpha)
strcat(out, "C ");
out += strlen(out);
}
strcat(out, "\n");
+
xf86DrvMsg(0, X_INFO, "%s", out2);
}
#else
-#define NV10EXAFallbackInfo(X,Y,Z,S,T) do { ; } while (0)
+#define print_fallback_info(...)
#endif
-Bool NV10EXACheckComposite(int op,
- PicturePtr pSrcPicture,
- PicturePtr pMaskPicture,
- PicturePtr pDstPicture)
+Bool
+NV10EXACheckComposite(int op, PicturePtr src, PicturePtr mask, PicturePtr dst)
{
-
- if (NV10Check_A8plusA8_Feasability(pSrcPicture, pMaskPicture, pDstPicture,
op)) {
- NV10EXAFallbackInfo("Hackelerating", op, pSrcPicture, pMaskPicture,
pDstPicture);
- return TRUE;
- }
-
- if (!NV10CheckPictOp(op)) {
- NV10EXAFallbackInfo("pictop", op, pSrcPicture, pMaskPicture,
pDstPicture);
+ if (!check_pict_op(op)) {
+ print_fallback_info("pictop", op, src, mask, dst);
return FALSE;
}
- if (!NV10CheckBuffer(pDstPicture)) {
- NV10EXAFallbackInfo("dst", op, pSrcPicture, pMaskPicture,
pDstPicture);
+
+ if (!check_render_target(dst)) {
+ print_fallback_info("dst", op, src, mask, dst);
return FALSE;
}
- if (!NV10CheckTexture(pSrcPicture)) {
- NV10EXAFallbackInfo("src", op, pSrcPicture, pMaskPicture,
pDstPicture);
+ if (!check_texture(src)) {
+ print_fallback_info("src", op, src, mask, dst);
return FALSE;
}
- if (pMaskPicture) {
- if (!NV10CheckTexture(pMaskPicture)) {
- NV10EXAFallbackInfo("mask", op, pSrcPicture,
- pMaskPicture, pDstPicture);
+ if (mask) {
+ if (!check_texture(mask)) {
+ print_fallback_info("mask", op, src,
+ mask, dst);
return FALSE;
}
- if (effective_component_alpha(pMaskPicture) &&
+ if (effective_component_alpha(mask) &&
needs_src(op) && needs_src_alpha(op)) {
- NV10EXAFallbackInfo("ca-mask", op, pSrcPicture,
- pMaskPicture, pDstPicture);
+ print_fallback_info("ca-mask", op, src,
+ mask, dst);
return FALSE;
}
}
- NV10EXAFallbackInfo("Accelerating", op, pSrcPicture, pMaskPicture,
pDstPicture);
+ print_fallback_info("Accelerating", op, src, mask, dst);
return TRUE;
}
-static void NV10SetTexture(NVPtr pNv, int unit, PicturePtr Pict, PixmapPtr
pixmap)
+static void
+setup_texture(NVPtr pNv, int unit, PicturePtr pict, PixmapPtr pixmap)
{
struct nouveau_channel *chan = pNv->chan;
struct nouveau_grobj *celsius = pNv->Nv3D;
struct nouveau_bo *bo = nouveau_pixmap_bo(pixmap);
unsigned delta = nouveau_pixmap_offset(pixmap);
- int log2w = log2i(Pict->pDrawable->width);
- int log2h = log2i(Pict->pDrawable->height);
- int w;
+ long w = pict->pDrawable->width,
+ h = pict->pDrawable->height;
unsigned int txfmt - (NV10TCL_TX_FORMAT_WRAP_T_CLAMP_TO_EDGE) |
- (NV10TCL_TX_FORMAT_WRAP_S_CLAMP_TO_EDGE) |
- (log2w<<20) |
- (log2h<<16) |
- (1<<12) | /* lod == 1 */
- 0x51 /* UNK */;
+ NV10TCL_TX_FORMAT_WRAP_T_CLAMP_TO_EDGE |
+ NV10TCL_TX_FORMAT_WRAP_S_CLAMP_TO_EDGE |
+ log2i(w) << 20 | log2i(h) << 16 |
+ 1 << 12 | /* lod == 1 */
+ get_tex_format(pict) |
+ 0x51 /* UNK */;
BEGIN_RING(chan, celsius, NV10TCL_TX_OFFSET(unit), 1);
OUT_RELOCl(chan, bo, delta, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD);
- /* if repeat is set we're always handling a 1x1 texture with ARGB/XRGB
destination,
- in that case we change the format to use the POT (swizzled) matching format */
- if (Pict->repeat != RepeatNone)
- {
- if (Pict->format == PICT_a8)
- txfmt |= 0x80; /* A8 */
- else if (Pict->format == PICT_r5g6b5 )
- txfmt |= 0x280; /* R5G6B5 */
- else
- txfmt |= 0x300; /* ARGB format */
- } else {
- if (pNv->Architecture == NV_ARCH_20 && Pict->format == PICT_a8)
- txfmt |= 0xd80;
- else
- txfmt |= NV10TexFormat(Pict->format);
- w = Pict->pDrawable->width;
- /* 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
- */
+ 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_RING(chan, celsius, NV10TCL_TX_NPOT_PITCH(unit), 1);
OUT_RING (chan, exaGetPixmapPitch(pixmap) << 16);
BEGIN_RING(chan, celsius, NV10TCL_TX_NPOT_SIZE(unit), 1);
- OUT_RING (chan, (w<<16) | Pict->pDrawable->height);
+ OUT_RING (chan, w << 16 | h);
}
BEGIN_RING(chan, celsius, NV10TCL_TX_FORMAT(unit), 1 );
@@ -403,214 +411,129 @@ static void NV10SetTexture(NVPtr pNv, int unit,
PicturePtr Pict, PixmapPtr pixma
OUT_RING (chan, NV10TCL_TX_ENABLE_ENABLE);
BEGIN_RING(chan, celsius, NV10TCL_TX_FILTER(unit), 1);
- if (Pict->filter == PictFilterNearest)
- OUT_RING (chan, (NV10TCL_TX_FILTER_MAGNIFY_NEAREST) |
- (NV10TCL_TX_FILTER_MINIFY_NEAREST));
+ if (pict->filter == PictFilterNearest)
+ OUT_RING(chan, (NV10TCL_TX_FILTER_MAGNIFY_NEAREST |
+ NV10TCL_TX_FILTER_MINIFY_NEAREST));
else
- OUT_RING (chan, (NV10TCL_TX_FILTER_MAGNIFY_LINEAR) |
- (NV10TCL_TX_FILTER_MINIFY_LINEAR));
-
- state.unit[unit].width = (float)pixmap->drawable.width;
- state.unit[unit].height = (float)pixmap->drawable.height;
- state.unit[unit].transform = Pict->transform;
+ OUT_RING(chan, (NV10TCL_TX_FILTER_MAGNIFY_LINEAR |
+ NV10TCL_TX_FILTER_MINIFY_LINEAR));
}
-static void NV10SetBuffer(NVPtr pNv, PicturePtr Pict, PixmapPtr pixmap)
+static void
+setup_render_target(NVPtr pNv, PicturePtr pict, PixmapPtr pixmap)
{
struct nouveau_channel *chan = pNv->chan;
struct nouveau_grobj *celsius = pNv->Nv3D;
struct nouveau_bo *bo = nouveau_pixmap_bo(pixmap);
unsigned delta = nouveau_pixmap_offset(pixmap);
- int i;
- int x = 0;
- int y = 0;
- int w = 2048;
- int h = 2048;
BEGIN_RING(chan, celsius, NV10TCL_RT_FORMAT, 4);
- if (state.is_a8_plus_a8) {
- /*A8 + A8 hack */
- OUT_RING (chan, NV10DstFormat(PICT_a8r8g8b8));
- } else {
- OUT_RING (chan, NV10DstFormat(Pict->format));
- }
-
- OUT_RING (chan, ((uint32_t)exaGetPixmapPitch(pixmap) << 16)
|(uint32_t)exaGetPixmapPitch(pixmap));
+ OUT_RING (chan, get_rt_format(pict));
+ OUT_RING (chan, (exaGetPixmapPitch(pixmap) << 16 |
+ exaGetPixmapPitch(pixmap)));
OUT_RELOCl(chan, bo, delta, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
OUT_RING (chan, 0);
+}
- BEGIN_RING(chan, celsius, NV10TCL_RT_HORIZ, 2);
- OUT_RING (chan, (w<<16)|x);
- OUT_RING (chan, (h<<16)|y);
- BEGIN_RING(chan, celsius, NV10TCL_VIEWPORT_CLIP_MODE, 1); /* clip_mode */
- OUT_RING (chan, 0);
- BEGIN_RING(chan, celsius, NV10TCL_VIEWPORT_CLIP_HORIZ(0), 1);
- OUT_RING (chan, ((w-1+x)<<16)|x|0x08000800);
- BEGIN_RING(chan, celsius, NV10TCL_VIEWPORT_CLIP_VERT(0), 1);
- OUT_RING (chan, ((h-1+y)<<16)|y|0x08000800);
+/*
+ * This can be a bit difficult to understand at first glance. Reg
+ * combiners are described here:
+ *
http://icps.u-strasbg.fr/~marchesin/perso/extensions/NV/register_combiners.html
+ *
+ * Single texturing setup, without honoring vertex colors (non default
+ * setup) is: Alpha RC 0 : a_0 * 1 + 0 * 0 RGB RC 0 : rgb_0 * 1 + 0 *
+ * 0 RC 1s are unused Final combiner uses default setup
+ *
+ * Default setup uses vertex rgb/alpha in place of 1s above, but we
+ * don't need that in 2D.
+ *
+ * Multi texturing setup, where we do TEX0 in TEX1 (masking) is:
+ * Alpha RC 0 : a_0 * a_1 + 0 * 0
+ * RGB RC0 : rgb_0 * a_1 + 0 * 0
+ * RC 1s are unused
+ * Final combiner uses default setup
+ */
- BEGIN_RING(chan, celsius, NV10TCL_PROJECTION_MATRIX(0), 16);
- for(i=0;i<16;i++)
- if (i/4==i%4)
- OUT_RINGf (chan, 1.0f);
- else
- OUT_RINGf (chan, 0.0f);
+/* Bind the combiner variable <input> to a constant 1. */
+#define RC_IN_ONE(input) \
+ (NV10TCL_RC_IN_RGB_##input##_INPUT_ZERO | \
+ NV10TCL_RC_IN_RGB_##input##_COMPONENT_USAGE_ALPHA | \
+ NV10TCL_RC_IN_RGB_##input##_MAPPING_UNSIGNED_INVERT_NV)
- BEGIN_RING(chan, celsius, NV10TCL_DEPTH_RANGE_NEAR, 2);
- OUT_RING (chan, 0);
-#if SCREEN_BPP == 32
- OUT_RINGf (chan, 16777216.0);
-#else
- OUT_RINGf (chan, 65536.0);
-#endif
- BEGIN_RING(chan, celsius, NV10TCL_VIEWPORT_SCALE_X, 4);
- OUT_RINGf (chan, -2048.0);
- OUT_RINGf (chan, -2048.0);
- OUT_RINGf (chan, 0);
- OUT_RING (chan, 0);
-}
+/* Bind the combiner variable <input> to the specified channel from
+ * the texture unit <unit>. */
+#define RC_IN_TEX(input, chan, unit) \
+ (NV10TCL_RC_IN_RGB_##input##_INPUT_TEXTURE##unit##_ARB | \
+ NV10TCL_RC_IN_RGB_##input##_COMPONENT_USAGE_##chan)
-#define RC_IN_ONE(chan, input) \
- (NV10TCL_RC_IN_##chan##_##input##_INPUT_ZERO \
- | NV10TCL_RC_IN_##chan##_##input##_COMPONENT_USAGE_##chan \
- | NV10TCL_RC_IN_##chan##_##input##_MAPPING_UNSIGNED_INVERT_NV)
+/* Bind the combiner variable <input> to the specified channel from
+ * the constant color <unit>. */
+#define RC_IN_COLOR(input, chan, unit) \
+ (NV10TCL_RC_IN_RGB_##input##_INPUT_CONSTANT_COLOR##unit##_NV | \
+ NV10TCL_RC_IN_RGB_##input##_COMPONENT_USAGE_##chan)
-static void NV10SetRegCombs(NVPtr pNv, PicturePtr src, PicturePtr mask)
+static void
+setup_combiners(NVPtr pNv, PicturePtr src, PicturePtr mask)
{
struct nouveau_channel *chan = pNv->chan;
struct nouveau_grobj *celsius = pNv->Nv3D;
- unsigned int rc0_in_alpha = 0, rc0_in_rgb = 0;
-
-/*This can be a bit difficult to understand at first glance.
-Reg combiners are described here:
-http://icps.u-strasbg.fr/~marchesin/perso/extensions/NV/register_combiners.html
-
-Single texturing setup, without honoring vertex colors (non default setup) is:
-Alpha RC 0 : a_0 * 1 + 0 * 0
-RGB RC 0 : rgb_0 * 1 + 0 * 0
-RC 1s are unused
-Final combiner uses default setup
-
-Default setup uses vertex rgb/alpha in place of 1s above, but we don't need
that in 2D.
-
-Multi texturing setup, where we do TEX0 in TEX1 (masking) is:
-Alpha RC 0 : a_0 * a_1 + 0 * 0
-RGB RC0 : rgb_0 * a_1 + 0 * 0
-RC 1s are unused
-Final combiner uses default setup
-
-*/
+ uint32_t rc_in_alpha = 0, rc_in_rgb = 0;
+
if (PICT_FORMAT_A(src->format))
- rc0_in_alpha |= NV10TCL_RC_IN_ALPHA_A_INPUT_TEXTURE0_ARB
- | NV10TCL_RC_IN_RGB_A_COMPONENT_USAGE_ALPHA;
+ rc_in_alpha |= RC_IN_TEX(A, ALPHA, 0);
else
- rc0_in_alpha |= RC_IN_ONE(ALPHA, A);
+ rc_in_alpha |= RC_IN_ONE(A);
if (mask && PICT_FORMAT_A(mask->format))
- rc0_in_alpha |= NV10TCL_RC_IN_RGB_B_INPUT_TEXTURE1_ARB
- | NV10TCL_RC_IN_RGB_B_COMPONENT_USAGE_ALPHA;
+ rc_in_alpha |= RC_IN_TEX(B, ALPHA, 1);
else
- rc0_in_alpha |= RC_IN_ONE(ALPHA, B);
+ rc_in_alpha |= RC_IN_ONE(B);
if (effective_component_alpha(mask)) {
- rc0_in_rgb |= NV10TCL_RC_IN_RGB_B_INPUT_TEXTURE1_ARB
- | NV10TCL_RC_IN_RGB_B_COMPONENT_USAGE_RGB;
-
if (!needs_src_alpha(pNv->alu)) {
- /*
- * The alpha channels won't be used for blending. Drop
+ /* 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))
- rc0_in_rgb |= NV10TCL_RC_IN_RGB_A_INPUT_TEXTURE0_ARB
- | NV10TCL_RC_IN_RGB_A_COMPONENT_USAGE_RGB;
-
+ rc_in_rgb |= RC_IN_TEX(A, RGB, 0);
} else {
- /*
- * The RGB channels won't be used for blending. Drop
+ /* The RGB channels won't be used for blending. Drop
* them.
* output_i = src_alpha * mask_i
*/
if (PICT_FORMAT_A(src->format))
- rc0_in_rgb |= NV10TCL_RC_IN_RGB_A_INPUT_TEXTURE0_ARB
- | NV10TCL_RC_IN_RGB_A_COMPONENT_USAGE_ALPHA;
+ rc_in_rgb |= RC_IN_TEX(A, ALPHA, 0);
else
- rc0_in_rgb |= RC_IN_ONE(RGB, A);
+ rc_in_rgb |= RC_IN_ONE(A);
}
+ rc_in_rgb |= RC_IN_TEX(B, RGB, 1);
+
} else {
if (PICT_FORMAT_RGB(src->format))
- rc0_in_rgb |= NV10TCL_RC_IN_RGB_A_INPUT_TEXTURE0_ARB
- | NV10TCL_RC_IN_RGB_A_COMPONENT_USAGE_RGB;
+ rc_in_rgb |= RC_IN_TEX(A, RGB, 0);
if (mask && PICT_FORMAT_A(mask->format))
- rc0_in_rgb |= NV10TCL_RC_IN_RGB_B_INPUT_TEXTURE1_ARB
- | NV10TCL_RC_IN_RGB_B_COMPONENT_USAGE_ALPHA;
+ rc_in_rgb |= RC_IN_TEX(B, ALPHA, 1);
else
- rc0_in_rgb |= RC_IN_ONE(RGB, B);
+ rc_in_rgb |= RC_IN_ONE(B);
}
BEGIN_RING(chan, celsius, NV10TCL_RC_IN_ALPHA(0), 1);
- OUT_RING (chan, rc0_in_alpha);
+ OUT_RING (chan, rc_in_alpha);
BEGIN_RING(chan, celsius, NV10TCL_RC_IN_RGB(0), 1);
- OUT_RING (chan, rc0_in_rgb);
+ OUT_RING (chan, rc_in_rgb);
}
-static void NV10SetRegCombs_A8plusA8(NVPtr pNv, int pass, int mask_out_bytes)
-{
- struct nouveau_channel *chan = pNv->chan;
- struct nouveau_grobj *celsius = pNv->Nv3D;
- unsigned int rc0_in_alpha = 0, rc0_in_rgb = 0;
- unsigned int color0 = 0, color1 = 0;
-
- if (pass == 1) {
- if (~mask_out_bytes & 1)
- rc0_in_alpha |= NV10TCL_RC_IN_ALPHA_A_INPUT_TEXTURE1_ARB
- | NV10TCL_RC_IN_ALPHA_A_COMPONENT_USAGE_ALPHA
- | RC_IN_ONE(ALPHA, B);
-
- if (~mask_out_bytes & 2)
- rc0_in_rgb |= NV10TCL_RC_IN_RGB_A_INPUT_TEXTURE0_ARB
- | NV10TCL_RC_IN_RGB_A_COMPONENT_USAGE_ALPHA
- | NV10TCL_RC_IN_RGB_B_INPUT_CONSTANT_COLOR0_NV
- | NV10TCL_RC_IN_RGB_B_COMPONENT_USAGE_RGB;
-
- color0 = 0x00ff0000; /*R = 1 G = 0 B = 0*/
- } else {
- if (~mask_out_bytes & 8)
- rc0_in_rgb |= NV10TCL_RC_IN_RGB_A_INPUT_TEXTURE0_ARB
- | NV10TCL_RC_IN_RGB_A_COMPONENT_USAGE_ALPHA
- | NV10TCL_RC_IN_RGB_B_INPUT_CONSTANT_COLOR0_NV
- | NV10TCL_RC_IN_RGB_B_COMPONENT_USAGE_RGB;
-
- color0 = 0x000000ff;
-
- if (~mask_out_bytes & 4)
- rc0_in_rgb |= NV10TCL_RC_IN_RGB_C_INPUT_TEXTURE1_ARB
- | NV10TCL_RC_IN_RGB_C_COMPONENT_USAGE_ALPHA
- | NV10TCL_RC_IN_RGB_D_INPUT_CONSTANT_COLOR1_NV
- | NV10TCL_RC_IN_RGB_D_COMPONENT_USAGE_RGB;
-
- color1 = 0x0000ff00; /*R = 0, G = 1, B = 0*/
- }
-
- BEGIN_RING(chan, celsius, NV10TCL_RC_IN_ALPHA(0), 1);
- OUT_RING (chan, rc0_in_alpha);
- BEGIN_RING(chan, celsius, NV10TCL_RC_IN_RGB(0), 1);
- OUT_RING (chan, rc0_in_rgb);
- BEGIN_RING(chan, celsius, NV10TCL_RC_COLOR(0), 2);
- OUT_RING (chan, color0);
- OUT_RING (chan, color1);
-}
-
-static void NV10SetPictOp(NVPtr pNv, int op)
+static void
+setup_blend_function(NVPtr pNv)
{
struct nouveau_channel *chan = pNv->chan;
struct nouveau_grobj *celsius = pNv->Nv3D;
- struct nv10_pictop *nv10_op = &NV10PictOp[op];
- int src_factor = nv10_op->src;
- int dst_factor = nv10_op->dst;
+ struct pict_op *op = &nv10_pict_op[pNv->alu];
+ int src_factor = op->src;
+ int dst_factor = op->dst;
if (src_factor == SF(ONE_MINUS_DST_ALPHA) &&
!PICT_FORMAT_A(pNv->pdpict->format))
@@ -644,280 +567,140 @@ NV10StateCompositeReemit(struct nouveau_channel *chan)
pNv->pspix, pNv->pmpix, pNv->pdpix);
}
-Bool NV10EXAPrepareComposite(int op,
- PicturePtr pSrcPicture,
- PicturePtr pMaskPicture,
- PicturePtr pDstPicture,
- PixmapPtr pSrc,
- PixmapPtr pMask,
- PixmapPtr pDst)
+Bool
+NV10EXAPrepareComposite(int op,
+ PicturePtr pict_src,
+ PicturePtr pict_mask,
+ PicturePtr pict_dst,
+ PixmapPtr src,
+ PixmapPtr mask,
+ PixmapPtr dst)
{
- ScrnInfoPtr pScrn = xf86Screens[pDst->drawable.pScreen->myNum];
+ ScrnInfoPtr pScrn = xf86Screens[dst->drawable.pScreen->myNum];
NVPtr pNv = NVPTR(pScrn);
struct nouveau_channel *chan = pNv->chan;
WAIT_RING(chan, 128);
- state.is_a8_plus_a8 = FALSE;
- state.have_mask=(pMaskPicture!=NULL);
-
pNv->alu = op;
- pNv->pspict = pSrcPicture;
- pNv->pmpict = pMaskPicture;
- pNv->pdpict = pDstPicture;
- pNv->pspix = pSrc;
- pNv->pmpix = pMask;
- pNv->pdpix = pDst;
+ pNv->pspict = pict_src;
+ pNv->pmpict = pict_mask;
+ pNv->pdpict = pict_dst;
+ pNv->pspix = src;
+ pNv->pmpix = mask;
+ pNv->pdpix = dst;
chan->flush_notify = NV10StateCompositeReemit;
- if (NV10Check_A8plusA8_Feasability(pSrcPicture, pMaskPicture, pDstPicture,
op)) {
- state.is_a8_plus_a8 = TRUE;
- NV10SetBuffer(pNv, pDstPicture, pDst);
- NV10SetPictOp(pNv, op);
- NV10SetTexture(pNv, 0, pSrcPicture, pSrc);
- NV10SetTexture(pNv, 1, pSrcPicture, pSrc);
- return TRUE;
- }
-
/* Set dst format */
- NV10SetBuffer(pNv, pDstPicture, pDst);
+ setup_render_target(pNv, pict_dst, dst);
/* Set src format */
- NV10SetTexture(pNv, 0, pSrcPicture, pSrc);
+ setup_texture(pNv, 0, pict_src, src);
/* Set mask format */
- if (pMaskPicture)
- NV10SetTexture(pNv, 1, pMaskPicture, pMask);
+ if (mask)
+ setup_texture(pNv, 1, pict_mask, mask);
- NV10SetRegCombs(pNv, pSrcPicture, pMaskPicture);
+ /* Set the register combiners up. */
+ setup_combiners(pNv, pict_src, pict_mask);
/* Set PictOp */
- NV10SetPictOp(pNv, op);
+ setup_blend_function(pNv);
return TRUE;
}
-static inline void NV10Vertex(NVPtr pNv, float vx, float vy, float tx, float
ty)
-{
- struct nouveau_channel *chan = pNv->chan;
- struct nouveau_grobj *celsius = pNv->Nv3D;
+#define QUAD(x, y, w, h) \
+ {{{ IntToxFixed(x), IntToxFixed(y), xFixed1 }}, \
+ {{ IntToxFixed(x + w), IntToxFixed(y), xFixed1 }}, \
+ {{ IntToxFixed(x + w), IntToxFixed(y + h), xFixed1 }}, \
+ {{ IntToxFixed(x), IntToxFixed(y + h), xFixed1 }}}
- BEGIN_RING(chan, celsius, NV10TCL_VERTEX_TX0_2F_S, 2);
- OUT_RINGf (chan, tx);
- OUT_RINGf (chan, ty);
- BEGIN_RING(chan, celsius, NV10TCL_VERTEX_POS_3F_X, 3);
- OUT_RINGf (chan, vx);
- OUT_RINGf (chan, vy);
- OUT_RINGf (chan, 0.f);
-}
+#define MAP(f, p, v, ...) do { \
+ int __i; \
+ for (__i = 0; __i < sizeof(v)/sizeof(*(v)); __i++) \
+ f(p, __i, v, ## __VA_ARGS__); \
+ } while (0);
-static inline void NV10MVertex(NVPtr pNv, float vx, float vy, float t0x, float
t0y, float t1x, float t1y)
+#define xFixedToFloat(v) \
+ ((float)xFixedToInt((v)) + ((float)xFixedFrac(v) / 65536.0))
+
+#define OUT_RINGi(chan, v, i) \
+ OUT_RINGf(chan, xFixedToFloat((v).vector[i]))
+
+static inline void
+emit_vertex(NVPtr pNv, int i, PictVector pos[],
+ PictVector tex0[], PictVector tex1[])
{
struct nouveau_channel *chan = pNv->chan;
struct nouveau_grobj *celsius = pNv->Nv3D;
BEGIN_RING(chan, celsius, NV10TCL_VERTEX_TX0_2F_S, 2);
- OUT_RINGf (chan, t0x);
- OUT_RINGf (chan, t0y);
- BEGIN_RING(chan, celsius, NV10TCL_VERTEX_TX1_2F_S, 2);
- OUT_RINGf (chan, t1x);
- OUT_RINGf (chan, t1y);
+ OUT_RINGi (chan, tex0[i], 0);
+ OUT_RINGi (chan, tex0[i], 1);
+
+ if (tex1) {
+ BEGIN_RING(chan, celsius, NV10TCL_VERTEX_TX1_2F_S, 2);
+ OUT_RINGi (chan, tex1[i], 0);
+ OUT_RINGi (chan, tex1[i], 1);
+ }
+
BEGIN_RING(chan, celsius, NV10TCL_VERTEX_POS_3F_X, 3);
- OUT_RINGf (chan, vx);
- OUT_RINGf (chan, vy);
- OUT_RINGf (chan, 0.f);
+ OUT_RINGi (chan, pos[i], 0);
+ OUT_RINGi (chan, pos[i], 1);
+ OUT_RINGf (chan, 0);
}
-#define xFixedToFloat(v) \
- ((float)xFixedToInt((v)) + ((float)xFixedFrac(v) / 65536.0))
-
-static void
-NV10EXATransformCoord(PictTransformPtr t, int x, int y, float sx, float sy,
- float *x_ret, float *y_ret)
+static inline void
+transform_vertex(PictTransformPtr t, int i, PictVector vs[])
{
- PictVector v;
-
- if (t) {
- v.vector[0] = IntToxFixed(x);
- v.vector[1] = IntToxFixed(y);
- v.vector[2] = xFixed1;
- PictureTransformPoint(t, &v);
- *x_ret = xFixedToFloat(v.vector[0]);
- *y_ret = xFixedToFloat(v.vector[1]);
- } else {
- *x_ret = (float)x;
- *y_ret = (float)y;
- }
+ if (t)
+ PictureTransformPoint(t, &vs[i]);
}
-
-void NV10EXAComposite(PixmapPtr pDst,
- int srcX,
- int srcY,
- int maskX,
- int maskY,
- int dstX,
- int dstY,
- int width,
- int height)
+void
+NV10EXAComposite(PixmapPtr pix_dst,
+ int srcX, int srcY,
+ int maskX, int maskY,
+ int dstX, int dstY,
+ int width, int height)
{
- ScrnInfoPtr pScrn = xf86Screens[pDst->drawable.pScreen->myNum];
+ ScrnInfoPtr pScrn = xf86Screens[pix_dst->drawable.pScreen->myNum];
NVPtr pNv = NVPTR(pScrn);
struct nouveau_channel *chan = pNv->chan;
struct nouveau_grobj *celsius = pNv->Nv3D;
- float sX0, sX1, sX2, sY0, sY1, sY2, sX3, sY3;
- float mX0, mX1, mX2, mY0, mY1, mY2, mX3, mY3;
+ 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_RING(chan, celsius, NV10TCL_VERTEX_BEGIN_END, 1);
OUT_RING (chan, NV10TCL_VERTEX_BEGIN_END_QUADS);
- NV10EXATransformCoord(state.unit[0].transform, srcX, srcY,
- state.unit[0].width,
- state.unit[0].height, &sX0, &sY0);
- NV10EXATransformCoord(state.unit[0].transform,
- srcX + width, srcY,
- state.unit[0].width,
- state.unit[0].height, &sX1, &sY1);
- NV10EXATransformCoord(state.unit[0].transform,
- srcX + width, srcY + height,
- state.unit[0].width,
- state.unit[0].height, &sX2, &sY2);
- NV10EXATransformCoord(state.unit[0].transform,
- srcX, srcY + height,
- state.unit[0].width,
- state.unit[0].height, &sX3, &sY3);
-
- if (state.is_a8_plus_a8) {
- /*We do A8 + A8 in 2-pass : setup the source texture as A8 twice,
- with different tex coords, do B and G on first pass
- Then setup again and do R and A on second pass
- */
- int part_pos_dX = 0;
- int part_pos_sX = 0;
- int mask_out_bytes = 0;
-
- part_pos_dX = (dstX & ~3) >> 2; /*we start at the 4byte boundary to
the left of the image */
- part_pos_sX = sX0 + (dstX & ~3) - dstX;
-
- /*xf86DrvMsg(0, X_INFO, "drawing - srcX %f dstX %d w %d\n", sX0,
dstX, width);*/
- for (; part_pos_dX <= (((dstX + width) & ~3) >> 2); part_pos_sX
+= 4, part_pos_dX++) {
- mask_out_bytes = 0;
- if (part_pos_dX == (dstX & ~3) >> 2) {
- /*we're slightly on the left of the image, bytes to mask out */
- /*xf86DrvMsg(0, X_INFO, "on left border...\n");*/
- switch (dstX - (dstX & ~3)) { /*mask out the extra pixels on the left
*/
- case 4:
- mask_out_bytes |= 1 << 0;
- case 3:
- mask_out_bytes |= 1 << 1;
- case 2:
- mask_out_bytes |= 1 << 2;
- case 1:
- mask_out_bytes |= 1 << 3;
- case 0:
- break;
- }
-
- /*mask out extra pixels on the right, in case the picture never touches an
alignment marker */
- switch (width + (dstX & 3)) {
- case 0:
- mask_out_bytes |= 1 << 3;
- case 1:
- mask_out_bytes |= 1 << 2;
- case 2:
- mask_out_bytes |= 1 << 1;
- case 3:
- mask_out_bytes |= 1 << 0;
- default:
- break;
- }
- } else if (part_pos_dX == (((dstX + width) & ~3) >> 2)) {
- /*xf86DrvMsg(0, X_INFO, "on right border...\n"); */
- switch (4 - ((dstX + width) & 3)) {
- case 4:
- mask_out_bytes |= 1 << 3;
- case 3:
- mask_out_bytes |= 1 << 2;
- case 2:
- mask_out_bytes |= 1 << 1;
- case 1:
- mask_out_bytes |= 1 << 0;
- case 0:
- break;
- }
- }
-
- /*Pass number 0*/
-
- NV10SetRegCombs_A8plusA8(pNv, 0, mask_out_bytes);
- BEGIN_RING(chan, celsius, NV10TCL_VERTEX_BEGIN_END, 1);
- OUT_RING (chan, NV10TCL_VERTEX_BEGIN_END_QUADS);
-
- NV10MVertex(pNv , part_pos_dX , dstY , part_pos_sX, sY0,
part_pos_sX + 1, sY0);
- NV10MVertex(pNv , part_pos_dX + 1, dstY , part_pos_sX, sY0,
part_pos_sX + 1, sY0);
- NV10MVertex(pNv , part_pos_dX + 1, dstY + height, part_pos_sX, sY2,
part_pos_sX + 1, sY2);
- NV10MVertex(pNv , part_pos_dX , dstY + height, part_pos_sX, sY2, part_pos_sX
+ 1, sY2);
-
- BEGIN_RING(chan, celsius, NV10TCL_VERTEX_BEGIN_END, 1);
- OUT_RING (chan, NV10TCL_VERTEX_BEGIN_END_STOP);
-
- /*Pass number 1*/
-
- NV10SetRegCombs_A8plusA8(pNv, 1, mask_out_bytes);
- BEGIN_RING(chan, celsius, NV10TCL_VERTEX_BEGIN_END, 1);
- OUT_RING (chan, NV10TCL_VERTEX_BEGIN_END_QUADS);
-
- NV10MVertex(pNv , part_pos_dX, dstY , part_pos_sX + 2, sY0,
part_pos_sX + 3, sY0);
- NV10MVertex(pNv , part_pos_dX + 1 , dstY , part_pos_sX + 2,
sY0, part_pos_sX + 3, sY0);
- NV10MVertex(pNv , part_pos_dX + 1 , dstY + height, part_pos_sX + 2, sY2,
part_pos_sX + 3, sY2);
- NV10MVertex(pNv , part_pos_dX, dstY + height, part_pos_sX + 2, sY2,
part_pos_sX + 3, sY2);
-
- BEGIN_RING(chan, celsius, NV10TCL_VERTEX_BEGIN_END, 1);
- OUT_RING (chan, NV10TCL_VERTEX_BEGIN_END_STOP);
-
- }
- } else if (state.have_mask) {
- NV10EXATransformCoord(state.unit[1].transform, maskX, maskY,
- state.unit[1].width,
- state.unit[1].height, &mX0, &mY0);
- NV10EXATransformCoord(state.unit[1].transform,
- maskX + width, maskY,
- state.unit[1].width,
- state.unit[1].height, &mX1, &mY1);
- NV10EXATransformCoord(state.unit[1].transform,
- maskX + width, maskY + height,
- state.unit[1].width,
- state.unit[1].height, &mX2, &mY2);
- NV10EXATransformCoord(state.unit[1].transform,
- maskX, maskY + height,
- state.unit[1].width,
- state.unit[1].height, &mX3, &mY3);
- NV10MVertex(pNv , dstX , dstY,sX0 , sY0 , mX0 , mY0);
- NV10MVertex(pNv , dstX + width , dstY,sX1 , sY1 , mX1 , mY1);
- NV10MVertex(pNv , dstX + width , dstY + height,sX2 , sY2 , mX2 , mY2);
- NV10MVertex(pNv , dstX , dstY + height,sX3 , sY3 , mX3 , mY3);
- } else {
- NV10Vertex(pNv , dstX , dstY , sX0 , sY0);
- NV10Vertex(pNv , dstX + width , dstY , sX1 , sY1);
- NV10Vertex(pNv , dstX + width , dstY + height , sX2 , sY2);
- NV10Vertex(pNv , dstX , dstY + height , sX3 , sY3);
- }
+ MAP(emit_vertex, pNv, dstq, srcq, mask ? maskq : NULL);
BEGIN_RING(chan, celsius, NV10TCL_VERTEX_BEGIN_END, 1);
OUT_RING (chan, NV10TCL_VERTEX_BEGIN_END_STOP);
}
-void NV10EXADoneComposite(PixmapPtr pDst)
+void
+NV10EXADoneComposite(PixmapPtr dst)
{
- ScrnInfoPtr pScrn = xf86Screens[pDst->drawable.pScreen->myNum];
+ ScrnInfoPtr pScrn = xf86Screens[dst->drawable.pScreen->myNum];
NVPtr pNv = NVPTR(pScrn);
struct nouveau_channel *chan = pNv->chan;
chan->flush_notify = NULL;
}
-Bool NVAccelInitNV10TCL(ScrnInfoPtr pScrn)
+Bool
+NVAccelInitNV10TCL(ScrnInfoPtr pScrn)
{
NVPtr pNv = NVPTR(pScrn);
struct nouveau_channel *chan = pNv->chan;
@@ -959,13 +742,16 @@ Bool NVAccelInitNV10TCL(ScrnInfoPtr pScrn)
OUT_RING (chan, 0);
BEGIN_RING(chan, celsius, NV10TCL_RT_HORIZ, 2);
- OUT_RING (chan, 0);
+ OUT_RING (chan, 2048 << 16 | 0);
+ OUT_RING (chan, 2048 << 16 | 0);
+
+ BEGIN_RING(chan, celsius, NV10TCL_VIEWPORT_CLIP_MODE, 1);
OUT_RING (chan, 0);
BEGIN_RING(chan, celsius, NV10TCL_VIEWPORT_CLIP_HORIZ(0), 1);
- OUT_RING (chan, (0x7ff<<16)|0x800);
+ OUT_RING (chan, 0x7ff << 16 | 0x800800);
BEGIN_RING(chan, celsius, NV10TCL_VIEWPORT_CLIP_VERT(0), 1);
- OUT_RING (chan, (0x7ff<<16)|0x800);
+ OUT_RING (chan, 0x7ff << 16 | 0x800800);
for (i = 1; i < 8; i++) {
BEGIN_RING(chan, celsius, NV10TCL_VIEWPORT_CLIP_HORIZ(i), 1);
@@ -1094,9 +880,9 @@ Bool NVAccelInitNV10TCL(ScrnInfoPtr pScrn)
BEGIN_RING(chan, celsius, NV10TCL_CULL_FACE_ENABLE, 1);
OUT_RING (chan, 0);
BEGIN_RING(chan, celsius, NV10TCL_TX_GEN_S(0), 8);
- for (i = 0; i < 8; i++) {
+ for (i = 0; i < 8; i++)
OUT_RING (chan, 0);
- }
+
BEGIN_RING(chan, celsius, NV10TCL_FOG_EQUATION_CONSTANT, 3);
OUT_RING (chan, 0x3fc00000); /* -1.50 */
OUT_RING (chan, 0xbdb8aa0a); /* -0.09 */
@@ -1116,6 +902,20 @@ Bool NVAccelInitNV10TCL(ScrnInfoPtr pScrn)
BEGIN_RING(chan, celsius, NV10TCL_COLOR_MASK, 1);
OUT_RING (chan, 0x01010101);
+ BEGIN_RING(chan, celsius, NV10TCL_PROJECTION_MATRIX(0), 16);
+ for(i = 0; i < 16; i++)
+ OUT_RINGf(chan, i/4 == i%4 ? 1.0 : 0.0);
+
+ BEGIN_RING(chan, celsius, NV10TCL_DEPTH_RANGE_NEAR, 2);
+ OUT_RING (chan, 0);
+ OUT_RINGf (chan, 65536.0);
+
+ BEGIN_RING(chan, celsius, NV10TCL_VIEWPORT_SCALE_X, 4);
+ OUT_RINGf (chan, -2048.0);
+ OUT_RINGf (chan, -2048.0);
+ OUT_RINGf (chan, 0);
+ OUT_RING (chan, 0);
+
/* Set vertex component */
BEGIN_RING(chan, celsius, NV10TCL_VERTEX_COL_4F_R, 4);
OUT_RINGf (chan, 1.0);
--
1.6.4.4
* Kill the A8+A8 hack. Recent enough X servers (>=1.7) fall back to
ARGB glyphs for drivers not supporting A8 render targets.
* Kill all the global state. It doesn't matter a lot yet but it might
if we get multicard working at some point.
* Other random clean-ups with no functional changes.
Some numbers from x11perf -aa10text -aa24text -comppixwin10 -comppixwin500:
* Before, with A glyphs:
12000000 trep @ 0.0025 msec (394000.0/sec): Char in 80-char aa line
(Charter 10)
3200000 trep @ 0.0119 msec ( 84300.0/sec): Char in 30-char aa line
(Charter 24)
8000000 trep @ 0.0036 msec (280000.0/sec): Composite 10x10 from pixmap to
window
60000 trep @ 0.4813 msec ( 2080.0/sec): Composite 500x500 from pixmap
to window
* After, ARGB glyphs:
20000000 trep @ 0.0015 msec (649000.0/sec): Char in 80-char aa line
(Charter 10)
4800000 trep @ 0.0057 msec (175000.0/sec): Char in 30-char aa line
(Charter 24)
8000000 trep @ 0.0036 msec (280000.0/sec): Composite 10x10 from pixmap to
window
60000 trep @ 0.4813 msec ( 2080.0/sec): Composite 500x500 from pixmap
to window
Signed-off-by: Francisco Jerez <currojerez at riseup.net>
---
src/nv10_exa.c | 921 ++++++++++++++++++++++----------------------------------
1 files changed, 361 insertions(+), 560 deletions(-)
diff --git a/src/nv10_exa.c b/src/nv10_exa.c
index fe86fd7..70f2fbc 100644
--- a/src/nv10_exa.c
+++ b/src/nv10_exa.c
@@ -2,6 +2,7 @@
* Copyright 2007 Stephane Marchesin
* Copyright 2007 Arthur Huillet
* Copyright 2007 Peter Winters
+ * Copyright 2009 Francisco Jerez
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the
"Software"),
@@ -28,24 +29,83 @@
#include "nv_include.h"
-typedef struct nv10_exa_state {
- Bool have_mask;
- Bool is_a8_plus_a8;
- struct {
- PictTransformPtr transform;
- float width;
- float height;
- } unit[2];
-} nv10_exa_state_t;
-static nv10_exa_state_t state;
+/* Texture/Render target formats. */
+static struct pict_format {
+ int exa;
+ int hw;
+} nv10_tex_format_pot[] = {
+ { PICT_a8, 0x80 },
+ { PICT_r5g6b5, 0x280 },
+ { PICT_x8r8g8b8, 0x300 },
+ { PICT_a8r8g8b8, 0x300 },
+ {},
+
+}, nv10_tex_format_rect[] = {
+ { PICT_a8, 0x980 },
+ { PICT_r5g6b5, 0x880 },
+ { PICT_x8r8g8b8, 0x900 },
+ { PICT_a8r8g8b8, 0x900 },
+ {},
+
+}, nv20_tex_format_rect[] = {
+ { PICT_a8, 0xd80 },
+ { PICT_r5g6b5, 0x880 },
+ { PICT_x8r8g8b8, 0x900 },
+ { PICT_a8r8g8b8, 0x900 },
+ {},
+
+}, nv10_rt_format[] = {
+ { PICT_r5g6b5, 0x103 },
+ { PICT_x8r8g8b8, 0x108 },
+ { PICT_a8r8g8b8, 0x108 },
+ {},
+};
+
+static int
+get_tex_format(PicturePtr pict)
+{
+ ScrnInfoPtr pScrn = xf86Screens[pict->pDrawable->pScreen->myNum];
+ NVPtr pNv = NVPTR(pScrn);
+
+ /* If repeat is set we're always handling a 1x1 texture with
+ * ARGB/XRGB destination, in that case we change the format to
+ * use the POT (swizzled) matching format.
+ */
+ struct pict_format *format + pict->repeat != RepeatNone ?
nv10_tex_format_pot :
+ pNv->Architecture == NV_ARCH_20 ? nv20_tex_format_rect :
+ nv10_tex_format_rect;
+
+ for (; format->hw; format++) {
+ if (format->exa == pict->format)
+ return format->hw;
+ }
+
+ return 0;
+}
+
+static int
+get_rt_format(PicturePtr pict)
+{
+ struct pict_format *format = nv10_rt_format;
+
+ for (; format->hw; format++) {
+ if (format->exa == pict->format)
+ return format->hw;
+ }
+
+ return 0;
+}
+/* Blending functions. */
#define SF(x) NV10TCL_BLEND_FUNC_SRC_##x
#define DF(x) NV10TCL_BLEND_FUNC_DST_##x
-static struct nv10_pictop {
+static struct pict_op {
int src;
int dst;
-} NV10PictOp [] = {
+
+} nv10_pict_op[] = {
{ SF(ZERO), DF(ZERO) }, /* Clear */
{ SF(ONE), DF(ZERO) }, /* Src */
{ SF(ZERO), DF(ONE) }, /* Dst */
@@ -61,131 +121,91 @@ static struct nv10_pictop {
{ SF(ONE), DF(ONE) }, /* Add */
};
-static inline bool needs_src_alpha(int op)
+static inline Bool
+needs_src_alpha(int op)
{
- return NV10PictOp[op].dst == DF(ONE_MINUS_SRC_ALPHA)
- || NV10PictOp[op].dst == DF(SRC_ALPHA);
+ return nv10_pict_op[op].dst == DF(ONE_MINUS_SRC_ALPHA)
+ || nv10_pict_op[op].dst == DF(SRC_ALPHA);
}
-static inline bool needs_src(int op)
+static inline Bool
+needs_src(int op)
{
- return NV10PictOp[op].src != DF(ZERO);
+ return nv10_pict_op[op].src != DF(ZERO);
}
-static inline bool effective_component_alpha(PicturePtr mask)
+static inline Bool
+effective_component_alpha(PicturePtr mask)
{
return mask && mask->componentAlpha &&
PICT_FORMAT_RGB(mask->format);
}
-static int NV10TexFormat(int ExaFormat)
-{
- struct {int exa;int hw;} tex_format[] - {
- {PICT_a8r8g8b8, 0x900},
- {PICT_x8r8g8b8, 0x900},
- {PICT_r5g6b5, 0x880}, /*this one was only tested with rendercheck*/
- /*{PICT_a1r5g5b5, NV10TCL_TX_FORMAT_FORMAT_R5G5B5A1},
- {PICT_a4r4g4b4, NV10TCL_TX_FORMAT_FORMAT_R4G4B4A4},*/
- {PICT_a8, 0x980}, /*this is a NV1x only format, corresponding NV2x is 0xD80,
we hack it in below*/
- };
-
- int i;
- for (i = 0; i < sizeof(tex_format) / sizeof(tex_format[0]); i++) {
- if (tex_format[i].exa == ExaFormat)
- return tex_format[i].hw;
- }
-
- return 0;
-}
-
-static int NV10DstFormat(int ExaFormat)
-{
- struct {int exa;int hw;} dst_format[] - {
- {PICT_a8r8g8b8, 0x108},
- {PICT_x8r8g8b8, 0x108},
- {PICT_r5g6b5, 0x103}
- };
-
- int i;
- for (i = 0; i < sizeof(dst_format) / sizeof(dst_format[0]); i++) {
- if (dst_format[i].exa == ExaFormat)
- return dst_format[i].hw;
- }
-
- return 0;
-}
-
-static Bool NV10CheckTexture(PicturePtr Picture)
+static Bool
+check_texture(PicturePtr pict)
{
int w, h;
- if (!Picture->pDrawable)
+ if (!pict->pDrawable)
NOUVEAU_FALLBACK("Solid and gradient pictures unsupported\n");
- w = Picture->pDrawable->width;
- h = Picture->pDrawable->height;
+ w = pict->pDrawable->width;
+ h = pict->pDrawable->height;
- if ((w > 2046) || (h > 2046))
+ if (w > 2046 || h > 2046)
NOUVEAU_FALLBACK("picture too large, %dx%d\n", w, h);
- if (!NV10TexFormat(Picture->format))
+ if (!get_tex_format(pict))
return FALSE;
- if (Picture->filter != PictFilterNearest && Picture->filter !=
PictFilterBilinear)
+
+ if (pict->filter != PictFilterNearest &&
+ pict->filter != PictFilterBilinear)
return FALSE;
- /* we cannot repeat on NV10 because NPOT textures do not support this.
unfortunately. */
- if (Picture->repeat != RepeatNone)
+
+ /* We cannot repeat on NV10 because NPOT textures do not
+ * support this. unfortunately. */
+ if (pict->repeat != RepeatNone)
/* we can repeat 1x1 textures */
if (!(w == 1 && h == 1))
return FALSE;
+
return TRUE;
}
-static Bool NV10CheckBuffer(PicturePtr Picture)
+static Bool
+check_render_target(PicturePtr pict)
{
- int w = Picture->pDrawable->width;
- int h = Picture->pDrawable->height;
+ int w = pict->pDrawable->width;
+ int h = pict->pDrawable->height;
- if ((w > 4096) || (h > 4096))
- return FALSE;
- if (!NV10DstFormat(Picture->format))
+ if (w > 4096 || h > 4096)
return FALSE;
- return TRUE;
-}
-static Bool NV10CheckPictOp(int op)
-{
- if (op >= PictOpSaturate) {
- /*we do no saturate, disjoint, conjoint, though we could do e.g.
DisjointClear which really is Clear */
+ if (!get_rt_format(pict))
return FALSE;
- }
+
return TRUE;
}
-/* Check if the current operation is a doable A8 + A8 */
-/* A8 destination is a special case, because we do it by having the card think
-it's ARGB. For now we support PictOpAdd which is the only important op for
this dst format,
-and without transformation or funny things.*/
-static Bool NV10Check_A8plusA8_Feasability(PicturePtr src, PicturePtr msk,
PicturePtr dst, int op)
+static Bool
+check_pict_op(int op)
{
-#if X_BYTE_ORDER == X_BIG_ENDIAN
- return FALSE;
-#endif
- if ((!msk) && (src->format == PICT_a8) && (dst->format
== PICT_a8) && (!src->transform) &&
- (op == PictOpAdd) && (src->repeat == RepeatNone)) {
- return TRUE;
- }
- return FALSE;
+ /* We do no saturate, disjoint, conjoint, though we
+ * could do e.g. DisjointClear which really is
+ * Clear. */
+ return op < PictOpSaturate;
}
#if 0
-static void NV10EXAFallbackInfo(char *reason, int op, PicturePtr pSrcPicture,
- PicturePtr pMaskPicture, PicturePtr pDstPicture)
+static void
+print_fallback_info(char *reason, int op, PicturePtr src, PicturePtr mask,
+ PicturePtr dst)
{
char out2[4096];
char *out = out2;
+
sprintf(out, "%s ", reason);
- out = out + strlen(out);
+ out += strlen(out);
+
switch (op) {
case PictOpClear:
sprintf(out, "PictOpClear ");
@@ -208,8 +228,9 @@ static void NV10EXAFallbackInfo(char *reason, int op,
PicturePtr pSrcPicture,
default:
sprintf(out, "PictOp%d ", op);
}
- out = out + strlen(out);
- switch (pSrcPicture->format) {
+ out += strlen(out);
+
+ switch (src->format) {
case PICT_a8r8g8b8:
sprintf(out, "A8R8G8B8 ");
break;
@@ -229,16 +250,18 @@ static void NV10EXAFallbackInfo(char *reason, int op,
PicturePtr pSrcPicture,
sprintf(out, "A1 ");
break;
default:
- sprintf(out, "%x ", pSrcPicture->format);
+ sprintf(out, "%x ", src->format);
}
out += strlen(out);
- sprintf(out, "(%dx%d) ", pSrcPicture->pDrawable->width,
pSrcPicture->pDrawable->height);
- if (pSrcPicture->repeat != RepeatNone)
+
+ sprintf(out, "(%dx%d) ", src->pDrawable->width,
+ src->pDrawable->height);
+ if (src->repeat != RepeatNone)
strcat(out, "R ");
strcat(out, "-> ");
out += strlen(out);
- switch (pDstPicture->format) {
+ switch (dst->format) {
case PICT_a8r8g8b8:
sprintf(out, "A8R8G8B8 ");
break;
@@ -258,17 +281,20 @@ static void NV10EXAFallbackInfo(char *reason, int op,
PicturePtr pSrcPicture,
sprintf(out, "A1 ");
break;
default:
- sprintf(out, "%x ", pDstPicture->format);
+ sprintf(out, "%x ", dst->format);
}
out += strlen(out);
- sprintf(out, "(%dx%d) ", pDstPicture->pDrawable->width,
pDstPicture->pDrawable->height);
- if (pDstPicture->repeat != RepeatNone)
+
+ sprintf(out, "(%dx%d) ", dst->pDrawable->width,
+ dst->pDrawable->height);
+ if (dst->repeat != RepeatNone)
strcat(out, "R ");
out += strlen(out);
- if (!pMaskPicture)
+
+ if (!mask)
sprintf(out, "& NONE");
else {
- switch (pMaskPicture->format) {
+ switch (mask->format) {
case PICT_a8r8g8b8:
sprintf(out, "& A8R8G8B8 ");
break;
@@ -285,116 +311,98 @@ static void NV10EXAFallbackInfo(char *reason, int op,
PicturePtr pSrcPicture,
sprintf(out, "& A1 ");
break;
default:
- sprintf(out, "& %x ", pMaskPicture->format);
+ sprintf(out, "& %x ", mask->format);
}
out += strlen(out);
- sprintf(out, "(%dx%d) ", pMaskPicture->pDrawable->width,
pMaskPicture->pDrawable->height);
- if (pMaskPicture->repeat != RepeatNone)
+
+ sprintf(out, "(%dx%d) ", mask->pDrawable->width,
+ mask->pDrawable->height);
+ if (mask->repeat != RepeatNone)
strcat(out, "R ");
- if (pMaskPicture->componentAlpha)
+ if (mask->componentAlpha)
strcat(out, "C ");
out += strlen(out);
}
strcat(out, "\n");
+
xf86DrvMsg(0, X_INFO, "%s", out2);
}
#else
-#define NV10EXAFallbackInfo(X,Y,Z,S,T) do { ; } while (0)
+#define print_fallback_info(...)
#endif
-Bool NV10EXACheckComposite(int op,
- PicturePtr pSrcPicture,
- PicturePtr pMaskPicture,
- PicturePtr pDstPicture)
+Bool
+NV10EXACheckComposite(int op, PicturePtr src, PicturePtr mask, PicturePtr dst)
{
-
- if (NV10Check_A8plusA8_Feasability(pSrcPicture, pMaskPicture, pDstPicture,
op)) {
- NV10EXAFallbackInfo("Hackelerating", op, pSrcPicture, pMaskPicture,
pDstPicture);
- return TRUE;
- }
-
- if (!NV10CheckPictOp(op)) {
- NV10EXAFallbackInfo("pictop", op, pSrcPicture, pMaskPicture,
pDstPicture);
+ if (!check_pict_op(op)) {
+ print_fallback_info("pictop", op, src, mask, dst);
return FALSE;
}
- if (!NV10CheckBuffer(pDstPicture)) {
- NV10EXAFallbackInfo("dst", op, pSrcPicture, pMaskPicture,
pDstPicture);
+
+ if (!check_render_target(dst)) {
+ print_fallback_info("dst", op, src, mask, dst);
return FALSE;
}
- if (!NV10CheckTexture(pSrcPicture)) {
- NV10EXAFallbackInfo("src", op, pSrcPicture, pMaskPicture,
pDstPicture);
+ if (!check_texture(src)) {
+ print_fallback_info("src", op, src, mask, dst);
return FALSE;
}
- if (pMaskPicture) {
- if (!NV10CheckTexture(pMaskPicture)) {
- NV10EXAFallbackInfo("mask", op, pSrcPicture,
- pMaskPicture, pDstPicture);
+ if (mask) {
+ if (!check_texture(mask)) {
+ print_fallback_info("mask", op, src,
+ mask, dst);
return FALSE;
}
- if (effective_component_alpha(pMaskPicture) &&
+ if (effective_component_alpha(mask) &&
needs_src(op) && needs_src_alpha(op)) {
- NV10EXAFallbackInfo("ca-mask", op, pSrcPicture,
- pMaskPicture, pDstPicture);
+ print_fallback_info("ca-mask", op, src,
+ mask, dst);
return FALSE;
}
}
- NV10EXAFallbackInfo("Accelerating", op, pSrcPicture, pMaskPicture,
pDstPicture);
+ print_fallback_info("Accelerating", op, src, mask, dst);
return TRUE;
}
-static Bool NV10SetTexture(NVPtr pNv, int unit, PicturePtr Pict, PixmapPtr
pixmap)
+static Bool
+setup_texture(NVPtr pNv, int unit, PicturePtr pict, PixmapPtr pixmap)
{
struct nouveau_channel *chan = pNv->chan;
struct nouveau_grobj *celsius = pNv->Nv3D;
struct nouveau_bo *bo = nouveau_pixmap_bo(pixmap);
unsigned delta = nouveau_pixmap_offset(pixmap);
- int log2w = log2i(Pict->pDrawable->width);
- int log2h = log2i(Pict->pDrawable->height);
- int w;
+ long w = pict->pDrawable->width,
+ h = pict->pDrawable->height;
unsigned int txfmt - (NV10TCL_TX_FORMAT_WRAP_T_CLAMP_TO_EDGE) |
- (NV10TCL_TX_FORMAT_WRAP_S_CLAMP_TO_EDGE) |
- (log2w<<20) |
- (log2h<<16) |
- (1<<12) | /* lod == 1 */
- 0x51 /* UNK */;
+ NV10TCL_TX_FORMAT_WRAP_T_CLAMP_TO_EDGE |
+ NV10TCL_TX_FORMAT_WRAP_S_CLAMP_TO_EDGE |
+ log2i(w) << 20 | log2i(h) << 16 |
+ 1 << 12 | /* lod == 1 */
+ get_tex_format(pict) |
+ 0x51 /* UNK */;
BEGIN_RING(chan, celsius, NV10TCL_TX_OFFSET(unit), 1);
if (OUT_RELOCl(chan, bo, delta, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD))
return FALSE;
- /* if repeat is set we're always handling a 1x1 texture with ARGB/XRGB
destination,
- in that case we change the format to use the POT (swizzled) matching format */
- if (Pict->repeat != RepeatNone)
- {
- if (Pict->format == PICT_a8)
- txfmt |= 0x80; /* A8 */
- else if (Pict->format == PICT_r5g6b5 )
- txfmt |= 0x280; /* R5G6B5 */
- else
- txfmt |= 0x300; /* ARGB format */
- } else {
- if (pNv->Architecture == NV_ARCH_20 && Pict->format == PICT_a8)
- txfmt |= 0xd80;
- else
- txfmt |= NV10TexFormat(Pict->format);
- w = Pict->pDrawable->width;
- /* 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
- */
+ 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_RING(chan, celsius, NV10TCL_TX_NPOT_PITCH(unit), 1);
OUT_RING (chan, exaGetPixmapPitch(pixmap) << 16);
BEGIN_RING(chan, celsius, NV10TCL_TX_NPOT_SIZE(unit), 1);
- OUT_RING (chan, (w<<16) | Pict->pDrawable->height);
+ OUT_RING (chan, w << 16 | h);
}
BEGIN_RING(chan, celsius, NV10TCL_TX_FORMAT(unit), 1 );
@@ -404,218 +412,135 @@ static Bool NV10SetTexture(NVPtr pNv, int unit,
PicturePtr Pict, PixmapPtr pixma
OUT_RING (chan, NV10TCL_TX_ENABLE_ENABLE);
BEGIN_RING(chan, celsius, NV10TCL_TX_FILTER(unit), 1);
- if (Pict->filter == PictFilterNearest)
- OUT_RING (chan, (NV10TCL_TX_FILTER_MAGNIFY_NEAREST) |
- (NV10TCL_TX_FILTER_MINIFY_NEAREST));
+ if (pict->filter == PictFilterNearest)
+ OUT_RING(chan, (NV10TCL_TX_FILTER_MAGNIFY_NEAREST |
+ NV10TCL_TX_FILTER_MINIFY_NEAREST));
else
- OUT_RING (chan, (NV10TCL_TX_FILTER_MAGNIFY_LINEAR) |
- (NV10TCL_TX_FILTER_MINIFY_LINEAR));
+ OUT_RING(chan, (NV10TCL_TX_FILTER_MAGNIFY_LINEAR |
+ NV10TCL_TX_FILTER_MINIFY_LINEAR));
- state.unit[unit].width = (float)pixmap->drawable.width;
- state.unit[unit].height = (float)pixmap->drawable.height;
- state.unit[unit].transform = Pict->transform;
return TRUE;
}
-static Bool NV10SetBuffer(NVPtr pNv, PicturePtr Pict, PixmapPtr pixmap)
+static Bool
+setup_render_target(NVPtr pNv, PicturePtr pict, PixmapPtr pixmap)
{
struct nouveau_channel *chan = pNv->chan;
struct nouveau_grobj *celsius = pNv->Nv3D;
struct nouveau_bo *bo = nouveau_pixmap_bo(pixmap);
unsigned delta = nouveau_pixmap_offset(pixmap);
- int i;
- int x = 0;
- int y = 0;
- int w = 2048;
- int h = 2048;
-
- BEGIN_RING(chan, celsius, NV10TCL_RT_FORMAT, 4);
- if (state.is_a8_plus_a8) {
- /*A8 + A8 hack */
- OUT_RING (chan, NV10DstFormat(PICT_a8r8g8b8));
- } else {
- OUT_RING (chan, NV10DstFormat(Pict->format));
- }
- OUT_RING (chan, ((uint32_t)exaGetPixmapPitch(pixmap) << 16)
|(uint32_t)exaGetPixmapPitch(pixmap));
+ BEGIN_RING(chan, celsius, NV10TCL_RT_FORMAT, 2);
+ OUT_RING (chan, get_rt_format(pict));
+ OUT_RING (chan, (exaGetPixmapPitch(pixmap) << 16 |
+ exaGetPixmapPitch(pixmap)));
+
+ BEGIN_RING(chan, celsius, NV10TCL_COLOR_OFFSET, 1);
if (OUT_RELOCl(chan, bo, delta, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR))
return FALSE;
- OUT_RING (chan, 0);
- BEGIN_RING(chan, celsius, NV10TCL_RT_HORIZ, 2);
- OUT_RING (chan, (w<<16)|x);
- OUT_RING (chan, (h<<16)|y);
- BEGIN_RING(chan, celsius, NV10TCL_VIEWPORT_CLIP_MODE, 1); /* clip_mode */
- OUT_RING (chan, 0);
- BEGIN_RING(chan, celsius, NV10TCL_VIEWPORT_CLIP_HORIZ(0), 1);
- OUT_RING (chan, ((w-1+x)<<16)|x|0x08000800);
- BEGIN_RING(chan, celsius, NV10TCL_VIEWPORT_CLIP_VERT(0), 1);
- OUT_RING (chan, ((h-1+y)<<16)|y|0x08000800);
+ return TRUE;
+}
- BEGIN_RING(chan, celsius, NV10TCL_PROJECTION_MATRIX(0), 16);
- for(i=0;i<16;i++)
- if (i/4==i%4)
- OUT_RINGf (chan, 1.0f);
- else
- OUT_RINGf (chan, 0.0f);
+/*
+ * This can be a bit difficult to understand at first glance. Reg
+ * combiners are described here:
+ *
http://icps.u-strasbg.fr/~marchesin/perso/extensions/NV/register_combiners.html
+ *
+ * Single texturing setup, without honoring vertex colors (non default
+ * setup) is: Alpha RC 0 : a_0 * 1 + 0 * 0 RGB RC 0 : rgb_0 * 1 + 0 *
+ * 0 RC 1s are unused Final combiner uses default setup
+ *
+ * Default setup uses vertex rgb/alpha in place of 1s above, but we
+ * don't need that in 2D.
+ *
+ * Multi texturing setup, where we do TEX0 in TEX1 (masking) is:
+ * Alpha RC 0 : a_0 * a_1 + 0 * 0
+ * RGB RC0 : rgb_0 * a_1 + 0 * 0
+ * RC 1s are unused
+ * Final combiner uses default setup
+ */
- BEGIN_RING(chan, celsius, NV10TCL_DEPTH_RANGE_NEAR, 2);
- OUT_RING (chan, 0);
-#if SCREEN_BPP == 32
- OUT_RINGf (chan, 16777216.0);
-#else
- OUT_RINGf (chan, 65536.0);
-#endif
- BEGIN_RING(chan, celsius, NV10TCL_VIEWPORT_SCALE_X, 4);
- OUT_RINGf (chan, -2048.0);
- OUT_RINGf (chan, -2048.0);
- OUT_RINGf (chan, 0);
- OUT_RING (chan, 0);
+/* Bind the combiner variable <input> to a constant 1. */
+#define RC_IN_ONE(input) \
+ (NV10TCL_RC_IN_RGB_##input##_INPUT_ZERO | \
+ NV10TCL_RC_IN_RGB_##input##_COMPONENT_USAGE_ALPHA | \
+ NV10TCL_RC_IN_RGB_##input##_MAPPING_UNSIGNED_INVERT_NV)
- return TRUE;
-}
+/* Bind the combiner variable <input> to the specified channel from
+ * the texture unit <unit>. */
+#define RC_IN_TEX(input, chan, unit) \
+ (NV10TCL_RC_IN_RGB_##input##_INPUT_TEXTURE##unit##_ARB | \
+ NV10TCL_RC_IN_RGB_##input##_COMPONENT_USAGE_##chan)
-#define RC_IN_ONE(chan, input) \
- (NV10TCL_RC_IN_##chan##_##input##_INPUT_ZERO \
- | NV10TCL_RC_IN_##chan##_##input##_COMPONENT_USAGE_##chan \
- | NV10TCL_RC_IN_##chan##_##input##_MAPPING_UNSIGNED_INVERT_NV)
+/* Bind the combiner variable <input> to the specified channel from
+ * the constant color <unit>. */
+#define RC_IN_COLOR(input, chan, unit) \
+ (NV10TCL_RC_IN_RGB_##input##_INPUT_CONSTANT_COLOR##unit##_NV | \
+ NV10TCL_RC_IN_RGB_##input##_COMPONENT_USAGE_##chan)
-static void NV10SetRegCombs(NVPtr pNv, PicturePtr src, PicturePtr mask)
+static void
+setup_combiners(NVPtr pNv, PicturePtr src, PicturePtr mask)
{
struct nouveau_channel *chan = pNv->chan;
struct nouveau_grobj *celsius = pNv->Nv3D;
- unsigned int rc0_in_alpha = 0, rc0_in_rgb = 0;
-
-/*This can be a bit difficult to understand at first glance.
-Reg combiners are described here:
-http://icps.u-strasbg.fr/~marchesin/perso/extensions/NV/register_combiners.html
-
-Single texturing setup, without honoring vertex colors (non default setup) is:
-Alpha RC 0 : a_0 * 1 + 0 * 0
-RGB RC 0 : rgb_0 * 1 + 0 * 0
-RC 1s are unused
-Final combiner uses default setup
-
-Default setup uses vertex rgb/alpha in place of 1s above, but we don't need
that in 2D.
-
-Multi texturing setup, where we do TEX0 in TEX1 (masking) is:
-Alpha RC 0 : a_0 * a_1 + 0 * 0
-RGB RC0 : rgb_0 * a_1 + 0 * 0
-RC 1s are unused
-Final combiner uses default setup
-
-*/
+ uint32_t rc_in_alpha = 0, rc_in_rgb = 0;
+
if (PICT_FORMAT_A(src->format))
- rc0_in_alpha |= NV10TCL_RC_IN_ALPHA_A_INPUT_TEXTURE0_ARB
- | NV10TCL_RC_IN_RGB_A_COMPONENT_USAGE_ALPHA;
+ rc_in_alpha |= RC_IN_TEX(A, ALPHA, 0);
else
- rc0_in_alpha |= RC_IN_ONE(ALPHA, A);
+ rc_in_alpha |= RC_IN_ONE(A);
if (mask && PICT_FORMAT_A(mask->format))
- rc0_in_alpha |= NV10TCL_RC_IN_RGB_B_INPUT_TEXTURE1_ARB
- | NV10TCL_RC_IN_RGB_B_COMPONENT_USAGE_ALPHA;
+ rc_in_alpha |= RC_IN_TEX(B, ALPHA, 1);
else
- rc0_in_alpha |= RC_IN_ONE(ALPHA, B);
+ rc_in_alpha |= RC_IN_ONE(B);
if (effective_component_alpha(mask)) {
- rc0_in_rgb |= NV10TCL_RC_IN_RGB_B_INPUT_TEXTURE1_ARB
- | NV10TCL_RC_IN_RGB_B_COMPONENT_USAGE_RGB;
-
if (!needs_src_alpha(pNv->alu)) {
- /*
- * The alpha channels won't be used for blending. Drop
+ /* 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))
- rc0_in_rgb |= NV10TCL_RC_IN_RGB_A_INPUT_TEXTURE0_ARB
- | NV10TCL_RC_IN_RGB_A_COMPONENT_USAGE_RGB;
-
+ rc_in_rgb |= RC_IN_TEX(A, RGB, 0);
} else {
- /*
- * The RGB channels won't be used for blending. Drop
+ /* The RGB channels won't be used for blending. Drop
* them.
* output_i = src_alpha * mask_i
*/
if (PICT_FORMAT_A(src->format))
- rc0_in_rgb |= NV10TCL_RC_IN_RGB_A_INPUT_TEXTURE0_ARB
- | NV10TCL_RC_IN_RGB_A_COMPONENT_USAGE_ALPHA;
+ rc_in_rgb |= RC_IN_TEX(A, ALPHA, 0);
else
- rc0_in_rgb |= RC_IN_ONE(RGB, A);
+ rc_in_rgb |= RC_IN_ONE(A);
}
+ rc_in_rgb |= RC_IN_TEX(B, RGB, 1);
+
} else {
if (PICT_FORMAT_RGB(src->format))
- rc0_in_rgb |= NV10TCL_RC_IN_RGB_A_INPUT_TEXTURE0_ARB
- | NV10TCL_RC_IN_RGB_A_COMPONENT_USAGE_RGB;
+ rc_in_rgb |= RC_IN_TEX(A, RGB, 0);
if (mask && PICT_FORMAT_A(mask->format))
- rc0_in_rgb |= NV10TCL_RC_IN_RGB_B_INPUT_TEXTURE1_ARB
- | NV10TCL_RC_IN_RGB_B_COMPONENT_USAGE_ALPHA;
+ rc_in_rgb |= RC_IN_TEX(B, ALPHA, 1);
else
- rc0_in_rgb |= RC_IN_ONE(RGB, B);
+ rc_in_rgb |= RC_IN_ONE(B);
}
BEGIN_RING(chan, celsius, NV10TCL_RC_IN_ALPHA(0), 1);
- OUT_RING (chan, rc0_in_alpha);
+ OUT_RING (chan, rc_in_alpha);
BEGIN_RING(chan, celsius, NV10TCL_RC_IN_RGB(0), 1);
- OUT_RING (chan, rc0_in_rgb);
+ OUT_RING (chan, rc_in_rgb);
}
-static void NV10SetRegCombs_A8plusA8(NVPtr pNv, int pass, int mask_out_bytes)
-{
- struct nouveau_channel *chan = pNv->chan;
- struct nouveau_grobj *celsius = pNv->Nv3D;
- unsigned int rc0_in_alpha = 0, rc0_in_rgb = 0;
- unsigned int color0 = 0, color1 = 0;
-
- if (pass == 1) {
- if (~mask_out_bytes & 1)
- rc0_in_alpha |= NV10TCL_RC_IN_ALPHA_A_INPUT_TEXTURE1_ARB
- | NV10TCL_RC_IN_ALPHA_A_COMPONENT_USAGE_ALPHA
- | RC_IN_ONE(ALPHA, B);
-
- if (~mask_out_bytes & 2)
- rc0_in_rgb |= NV10TCL_RC_IN_RGB_A_INPUT_TEXTURE0_ARB
- | NV10TCL_RC_IN_RGB_A_COMPONENT_USAGE_ALPHA
- | NV10TCL_RC_IN_RGB_B_INPUT_CONSTANT_COLOR0_NV
- | NV10TCL_RC_IN_RGB_B_COMPONENT_USAGE_RGB;
-
- color0 = 0x00ff0000; /*R = 1 G = 0 B = 0*/
- } else {
- if (~mask_out_bytes & 8)
- rc0_in_rgb |= NV10TCL_RC_IN_RGB_A_INPUT_TEXTURE0_ARB
- | NV10TCL_RC_IN_RGB_A_COMPONENT_USAGE_ALPHA
- | NV10TCL_RC_IN_RGB_B_INPUT_CONSTANT_COLOR0_NV
- | NV10TCL_RC_IN_RGB_B_COMPONENT_USAGE_RGB;
-
- color0 = 0x000000ff;
-
- if (~mask_out_bytes & 4)
- rc0_in_rgb |= NV10TCL_RC_IN_RGB_C_INPUT_TEXTURE1_ARB
- | NV10TCL_RC_IN_RGB_C_COMPONENT_USAGE_ALPHA
- | NV10TCL_RC_IN_RGB_D_INPUT_CONSTANT_COLOR1_NV
- | NV10TCL_RC_IN_RGB_D_COMPONENT_USAGE_RGB;
-
- color1 = 0x0000ff00; /*R = 0, G = 1, B = 0*/
- }
-
- BEGIN_RING(chan, celsius, NV10TCL_RC_IN_ALPHA(0), 1);
- OUT_RING (chan, rc0_in_alpha);
- BEGIN_RING(chan, celsius, NV10TCL_RC_IN_RGB(0), 1);
- OUT_RING (chan, rc0_in_rgb);
- BEGIN_RING(chan, celsius, NV10TCL_RC_COLOR(0), 2);
- OUT_RING (chan, color0);
- OUT_RING (chan, color1);
-}
-
-static void NV10SetPictOp(NVPtr pNv, int op)
+static void
+setup_blend_function(NVPtr pNv)
{
struct nouveau_channel *chan = pNv->chan;
struct nouveau_grobj *celsius = pNv->Nv3D;
- struct nv10_pictop *nv10_op = &NV10PictOp[op];
- int src_factor = nv10_op->src;
- int dst_factor = nv10_op->dst;
+ struct pict_op *op = &nv10_pict_op[pNv->alu];
+ int src_factor = op->src;
+ int dst_factor = op->dst;
if (src_factor == SF(ONE_MINUS_DST_ALPHA) &&
!PICT_FORMAT_A(pNv->pdpict->format))
@@ -649,294 +574,150 @@ NV10StateCompositeReemit(struct nouveau_channel *chan)
pNv->pspix, pNv->pmpix, pNv->pdpix);
}
-Bool NV10EXAPrepareComposite(int op,
- PicturePtr pSrcPicture,
- PicturePtr pMaskPicture,
- PicturePtr pDstPicture,
- PixmapPtr pSrc,
- PixmapPtr pMask,
- PixmapPtr pDst)
+Bool
+NV10EXAPrepareComposite(int op,
+ PicturePtr pict_src,
+ PicturePtr pict_mask,
+ PicturePtr pict_dst,
+ PixmapPtr src,
+ PixmapPtr mask,
+ PixmapPtr dst)
{
- ScrnInfoPtr pScrn = xf86Screens[pDst->drawable.pScreen->myNum];
+ ScrnInfoPtr pScrn = xf86Screens[dst->drawable.pScreen->myNum];
NVPtr pNv = NVPTR(pScrn);
struct nouveau_channel *chan = pNv->chan;
if (MARK_RING(chan, 128, 3))
return FALSE;
- state.is_a8_plus_a8 = FALSE;
- state.have_mask=(pMaskPicture!=NULL);
-
pNv->alu = op;
- pNv->pspict = pSrcPicture;
- pNv->pmpict = pMaskPicture;
- pNv->pdpict = pDstPicture;
- pNv->pspix = pSrc;
- pNv->pmpix = pMask;
- pNv->pdpix = pDst;
- chan->flush_notify = NV10StateCompositeReemit;
-
- if (NV10Check_A8plusA8_Feasability(pSrcPicture, pMaskPicture, pDstPicture,
op)) {
- state.is_a8_plus_a8 = TRUE;
- if (!NV10SetBuffer(pNv, pDstPicture, pDst) ||
- !NV10SetTexture(pNv, 0, pSrcPicture, pSrc) ||
- !NV10SetTexture(pNv, 1, pSrcPicture, pSrc)) {
- MARK_UNDO(chan);
- return FALSE;
- }
- NV10SetPictOp(pNv, op);
- return TRUE;
- }
+ pNv->pspict = pict_src;
+ pNv->pmpict = pict_mask;
+ pNv->pdpict = pict_dst;
+ pNv->pspix = src;
+ pNv->pmpix = mask;
+ pNv->pdpix = dst;
/* Set dst format */
- if (!NV10SetBuffer(pNv, pDstPicture, pDst)) {
- MARK_UNDO(chan);
- return FALSE;
- }
+ if (!setup_render_target(pNv, pict_dst, dst))
+ goto fail;
/* Set src format */
- if (!NV10SetTexture(pNv, 0, pSrcPicture, pSrc)) {
- MARK_UNDO(chan);
- return FALSE;
- }
+ if (!setup_texture(pNv, 0, pict_src, src))
+ goto fail;
/* Set mask format */
- if (pMaskPicture) {
- if (!NV10SetTexture(pNv, 1, pMaskPicture, pMask)) {
- MARK_UNDO(chan);
- return FALSE;
- }
- }
+ if (mask &&
+ !setup_texture(pNv, 1, pict_mask, mask))
+ goto fail;
- NV10SetRegCombs(pNv, pSrcPicture, pMaskPicture);
+ /* Set the register combiners up. */
+ setup_combiners(pNv, pict_src, pict_mask);
/* Set PictOp */
- NV10SetPictOp(pNv, op);
+ setup_blend_function(pNv);
+
+ chan->flush_notify = NV10StateCompositeReemit;
return TRUE;
-}
-static inline void NV10Vertex(NVPtr pNv, float vx, float vy, float tx, float
ty)
-{
- struct nouveau_channel *chan = pNv->chan;
- struct nouveau_grobj *celsius = pNv->Nv3D;
+fail:
+ MARK_UNDO(chan);
- BEGIN_RING(chan, celsius, NV10TCL_VERTEX_TX0_2F_S, 2);
- OUT_RINGf (chan, tx);
- OUT_RINGf (chan, ty);
- BEGIN_RING(chan, celsius, NV10TCL_VERTEX_POS_3F_X, 3);
- OUT_RINGf (chan, vx);
- OUT_RINGf (chan, vy);
- OUT_RINGf (chan, 0.f);
+ return FALSE;
}
-static inline void NV10MVertex(NVPtr pNv, float vx, float vy, float t0x, float
t0y, float t1x, float t1y)
+#define QUAD(x, y, w, h) \
+ {{{ IntToxFixed(x), IntToxFixed(y), xFixed1 }}, \
+ {{ IntToxFixed(x + w), IntToxFixed(y), xFixed1 }}, \
+ {{ IntToxFixed(x + w), IntToxFixed(y + h), xFixed1 }}, \
+ {{ IntToxFixed(x), IntToxFixed(y + h), xFixed1 }}}
+
+#define MAP(f, p, v, ...) do { \
+ int __i; \
+ for (__i = 0; __i < sizeof(v)/sizeof((v)[0]); __i++) \
+ f(p, __i, v, ## __VA_ARGS__); \
+ } while (0);
+
+#define xFixedToFloat(v) \
+ ((float)xFixedToInt((v)) + ((float)xFixedFrac(v) / 65536.0))
+
+#define OUT_RINGi(chan, v, i) \
+ OUT_RINGf(chan, xFixedToFloat((v).vector[i]))
+
+static inline void
+emit_vertex(NVPtr pNv, int i, PictVector pos[],
+ PictVector tex0[], PictVector tex1[])
{
struct nouveau_channel *chan = pNv->chan;
struct nouveau_grobj *celsius = pNv->Nv3D;
BEGIN_RING(chan, celsius, NV10TCL_VERTEX_TX0_2F_S, 2);
- OUT_RINGf (chan, t0x);
- OUT_RINGf (chan, t0y);
- BEGIN_RING(chan, celsius, NV10TCL_VERTEX_TX1_2F_S, 2);
- OUT_RINGf (chan, t1x);
- OUT_RINGf (chan, t1y);
+ OUT_RINGi (chan, tex0[i], 0);
+ OUT_RINGi (chan, tex0[i], 1);
+
+ if (tex1) {
+ BEGIN_RING(chan, celsius, NV10TCL_VERTEX_TX1_2F_S, 2);
+ OUT_RINGi (chan, tex1[i], 0);
+ OUT_RINGi (chan, tex1[i], 1);
+ }
+
BEGIN_RING(chan, celsius, NV10TCL_VERTEX_POS_3F_X, 3);
- OUT_RINGf (chan, vx);
- OUT_RINGf (chan, vy);
- OUT_RINGf (chan, 0.f);
+ OUT_RINGi (chan, pos[i], 0);
+ OUT_RINGi (chan, pos[i], 1);
+ OUT_RINGf (chan, 0);
}
-#define xFixedToFloat(v) \
- ((float)xFixedToInt((v)) + ((float)xFixedFrac(v) / 65536.0))
-
-static void
-NV10EXATransformCoord(PictTransformPtr t, int x, int y, float sx, float sy,
- float *x_ret, float *y_ret)
+static inline void
+transform_vertex(PictTransformPtr t, int i, PictVector vs[])
{
- PictVector v;
-
- if (t) {
- v.vector[0] = IntToxFixed(x);
- v.vector[1] = IntToxFixed(y);
- v.vector[2] = xFixed1;
- PictureTransformPoint(t, &v);
- *x_ret = xFixedToFloat(v.vector[0]);
- *y_ret = xFixedToFloat(v.vector[1]);
- } else {
- *x_ret = (float)x;
- *y_ret = (float)y;
- }
+ if (t)
+ PictureTransformPoint(t, &vs[i]);
}
-
-void NV10EXAComposite(PixmapPtr pDst,
- int srcX,
- int srcY,
- int maskX,
- int maskY,
- int dstX,
- int dstY,
- int width,
- int height)
+void
+NV10EXAComposite(PixmapPtr pix_dst,
+ int srcX, int srcY,
+ int maskX, int maskY,
+ int dstX, int dstY,
+ int width, int height)
{
- ScrnInfoPtr pScrn = xf86Screens[pDst->drawable.pScreen->myNum];
+ ScrnInfoPtr pScrn = xf86Screens[pix_dst->drawable.pScreen->myNum];
NVPtr pNv = NVPTR(pScrn);
struct nouveau_channel *chan = pNv->chan;
struct nouveau_grobj *celsius = pNv->Nv3D;
- float sX0, sX1, sX2, sY0, sY1, sY2, sX3, sY3;
- float mX0, mX1, mX2, mY0, mY1, mY2, mX3, mY3;
+ 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_RING(chan, celsius, NV10TCL_VERTEX_BEGIN_END, 1);
OUT_RING (chan, NV10TCL_VERTEX_BEGIN_END_QUADS);
- NV10EXATransformCoord(state.unit[0].transform, srcX, srcY,
- state.unit[0].width,
- state.unit[0].height, &sX0, &sY0);
- NV10EXATransformCoord(state.unit[0].transform,
- srcX + width, srcY,
- state.unit[0].width,
- state.unit[0].height, &sX1, &sY1);
- NV10EXATransformCoord(state.unit[0].transform,
- srcX + width, srcY + height,
- state.unit[0].width,
- state.unit[0].height, &sX2, &sY2);
- NV10EXATransformCoord(state.unit[0].transform,
- srcX, srcY + height,
- state.unit[0].width,
- state.unit[0].height, &sX3, &sY3);
-
- if (state.is_a8_plus_a8) {
- /*We do A8 + A8 in 2-pass : setup the source texture as A8 twice,
- with different tex coords, do B and G on first pass
- Then setup again and do R and A on second pass
- */
- int part_pos_dX = 0;
- int part_pos_sX = 0;
- int mask_out_bytes = 0;
-
- part_pos_dX = (dstX & ~3) >> 2; /*we start at the 4byte boundary to
the left of the image */
- part_pos_sX = sX0 + (dstX & ~3) - dstX;
-
- /*xf86DrvMsg(0, X_INFO, "drawing - srcX %f dstX %d w %d\n", sX0,
dstX, width);*/
- for (; part_pos_dX <= (((dstX + width) & ~3) >> 2); part_pos_sX
+= 4, part_pos_dX++) {
- mask_out_bytes = 0;
- if (part_pos_dX == (dstX & ~3) >> 2) {
- /*we're slightly on the left of the image, bytes to mask out */
- /*xf86DrvMsg(0, X_INFO, "on left border...\n");*/
- switch (dstX - (dstX & ~3)) { /*mask out the extra pixels on the left
*/
- case 4:
- mask_out_bytes |= 1 << 0;
- case 3:
- mask_out_bytes |= 1 << 1;
- case 2:
- mask_out_bytes |= 1 << 2;
- case 1:
- mask_out_bytes |= 1 << 3;
- case 0:
- break;
- }
-
- /*mask out extra pixels on the right, in case the picture never touches an
alignment marker */
- switch (width + (dstX & 3)) {
- case 0:
- mask_out_bytes |= 1 << 3;
- case 1:
- mask_out_bytes |= 1 << 2;
- case 2:
- mask_out_bytes |= 1 << 1;
- case 3:
- mask_out_bytes |= 1 << 0;
- default:
- break;
- }
- } else if (part_pos_dX == (((dstX + width) & ~3) >> 2)) {
- /*xf86DrvMsg(0, X_INFO, "on right border...\n"); */
- switch (4 - ((dstX + width) & 3)) {
- case 4:
- mask_out_bytes |= 1 << 3;
- case 3:
- mask_out_bytes |= 1 << 2;
- case 2:
- mask_out_bytes |= 1 << 1;
- case 1:
- mask_out_bytes |= 1 << 0;
- case 0:
- break;
- }
- }
-
- /*Pass number 0*/
-
- NV10SetRegCombs_A8plusA8(pNv, 0, mask_out_bytes);
- BEGIN_RING(chan, celsius, NV10TCL_VERTEX_BEGIN_END, 1);
- OUT_RING (chan, NV10TCL_VERTEX_BEGIN_END_QUADS);
-
- NV10MVertex(pNv , part_pos_dX , dstY , part_pos_sX, sY0,
part_pos_sX + 1, sY0);
- NV10MVertex(pNv , part_pos_dX + 1, dstY , part_pos_sX, sY0,
part_pos_sX + 1, sY0);
- NV10MVertex(pNv , part_pos_dX + 1, dstY + height, part_pos_sX, sY2,
part_pos_sX + 1, sY2);
- NV10MVertex(pNv , part_pos_dX , dstY + height, part_pos_sX, sY2, part_pos_sX
+ 1, sY2);
-
- BEGIN_RING(chan, celsius, NV10TCL_VERTEX_BEGIN_END, 1);
- OUT_RING (chan, NV10TCL_VERTEX_BEGIN_END_STOP);
-
- /*Pass number 1*/
-
- NV10SetRegCombs_A8plusA8(pNv, 1, mask_out_bytes);
- BEGIN_RING(chan, celsius, NV10TCL_VERTEX_BEGIN_END, 1);
- OUT_RING (chan, NV10TCL_VERTEX_BEGIN_END_QUADS);
-
- NV10MVertex(pNv , part_pos_dX, dstY , part_pos_sX + 2, sY0,
part_pos_sX + 3, sY0);
- NV10MVertex(pNv , part_pos_dX + 1 , dstY , part_pos_sX + 2,
sY0, part_pos_sX + 3, sY0);
- NV10MVertex(pNv , part_pos_dX + 1 , dstY + height, part_pos_sX + 2, sY2,
part_pos_sX + 3, sY2);
- NV10MVertex(pNv , part_pos_dX, dstY + height, part_pos_sX + 2, sY2,
part_pos_sX + 3, sY2);
-
- BEGIN_RING(chan, celsius, NV10TCL_VERTEX_BEGIN_END, 1);
- OUT_RING (chan, NV10TCL_VERTEX_BEGIN_END_STOP);
-
- }
- } else if (state.have_mask) {
- NV10EXATransformCoord(state.unit[1].transform, maskX, maskY,
- state.unit[1].width,
- state.unit[1].height, &mX0, &mY0);
- NV10EXATransformCoord(state.unit[1].transform,
- maskX + width, maskY,
- state.unit[1].width,
- state.unit[1].height, &mX1, &mY1);
- NV10EXATransformCoord(state.unit[1].transform,
- maskX + width, maskY + height,
- state.unit[1].width,
- state.unit[1].height, &mX2, &mY2);
- NV10EXATransformCoord(state.unit[1].transform,
- maskX, maskY + height,
- state.unit[1].width,
- state.unit[1].height, &mX3, &mY3);
- NV10MVertex(pNv , dstX , dstY,sX0 , sY0 , mX0 , mY0);
- NV10MVertex(pNv , dstX + width , dstY,sX1 , sY1 , mX1 , mY1);
- NV10MVertex(pNv , dstX + width , dstY + height,sX2 , sY2 , mX2 , mY2);
- NV10MVertex(pNv , dstX , dstY + height,sX3 , sY3 , mX3 , mY3);
- } else {
- NV10Vertex(pNv , dstX , dstY , sX0 , sY0);
- NV10Vertex(pNv , dstX + width , dstY , sX1 , sY1);
- NV10Vertex(pNv , dstX + width , dstY + height , sX2 , sY2);
- NV10Vertex(pNv , dstX , dstY + height , sX3 , sY3);
- }
+ MAP(emit_vertex, pNv, dstq, srcq, mask ? maskq : NULL);
BEGIN_RING(chan, celsius, NV10TCL_VERTEX_BEGIN_END, 1);
OUT_RING (chan, NV10TCL_VERTEX_BEGIN_END_STOP);
}
-void NV10EXADoneComposite(PixmapPtr pDst)
+void
+NV10EXADoneComposite(PixmapPtr dst)
{
- ScrnInfoPtr pScrn = xf86Screens[pDst->drawable.pScreen->myNum];
+ ScrnInfoPtr pScrn = xf86Screens[dst->drawable.pScreen->myNum];
NVPtr pNv = NVPTR(pScrn);
struct nouveau_channel *chan = pNv->chan;
chan->flush_notify = NULL;
}
-Bool NVAccelInitNV10TCL(ScrnInfoPtr pScrn)
+Bool
+NVAccelInitNV10TCL(ScrnInfoPtr pScrn)
{
NVPtr pNv = NVPTR(pScrn);
struct nouveau_channel *chan = pNv->chan;
@@ -978,13 +759,19 @@ Bool NVAccelInitNV10TCL(ScrnInfoPtr pScrn)
OUT_RING (chan, 0);
BEGIN_RING(chan, celsius, NV10TCL_RT_HORIZ, 2);
+ OUT_RING (chan, 2048 << 16 | 0);
+ OUT_RING (chan, 2048 << 16 | 0);
+
+ BEGIN_RING(chan, celsius, NV10TCL_ZETA_OFFSET, 1);
OUT_RING (chan, 0);
+
+ BEGIN_RING(chan, celsius, NV10TCL_VIEWPORT_CLIP_MODE, 1);
OUT_RING (chan, 0);
BEGIN_RING(chan, celsius, NV10TCL_VIEWPORT_CLIP_HORIZ(0), 1);
- OUT_RING (chan, (0x7ff<<16)|0x800);
+ OUT_RING (chan, 0x7ff << 16 | 0x800800);
BEGIN_RING(chan, celsius, NV10TCL_VIEWPORT_CLIP_VERT(0), 1);
- OUT_RING (chan, (0x7ff<<16)|0x800);
+ OUT_RING (chan, 0x7ff << 16 | 0x800800);
for (i = 1; i < 8; i++) {
BEGIN_RING(chan, celsius, NV10TCL_VIEWPORT_CLIP_HORIZ(i), 1);
@@ -1113,9 +900,9 @@ Bool NVAccelInitNV10TCL(ScrnInfoPtr pScrn)
BEGIN_RING(chan, celsius, NV10TCL_CULL_FACE_ENABLE, 1);
OUT_RING (chan, 0);
BEGIN_RING(chan, celsius, NV10TCL_TX_GEN_S(0), 8);
- for (i = 0; i < 8; i++) {
+ for (i = 0; i < 8; i++)
OUT_RING (chan, 0);
- }
+
BEGIN_RING(chan, celsius, NV10TCL_FOG_EQUATION_CONSTANT, 3);
OUT_RING (chan, 0x3fc00000); /* -1.50 */
OUT_RING (chan, 0xbdb8aa0a); /* -0.09 */
@@ -1135,6 +922,20 @@ Bool NVAccelInitNV10TCL(ScrnInfoPtr pScrn)
BEGIN_RING(chan, celsius, NV10TCL_COLOR_MASK, 1);
OUT_RING (chan, 0x01010101);
+ BEGIN_RING(chan, celsius, NV10TCL_PROJECTION_MATRIX(0), 16);
+ for(i = 0; i < 16; i++)
+ OUT_RINGf(chan, i/4 == i%4 ? 1.0 : 0.0);
+
+ BEGIN_RING(chan, celsius, NV10TCL_DEPTH_RANGE_NEAR, 2);
+ OUT_RING (chan, 0);
+ OUT_RINGf (chan, 65536.0);
+
+ BEGIN_RING(chan, celsius, NV10TCL_VIEWPORT_SCALE_X, 4);
+ OUT_RINGf (chan, -2048.0);
+ OUT_RINGf (chan, -2048.0);
+ OUT_RINGf (chan, 0);
+ OUT_RING (chan, 0);
+
/* Set vertex component */
BEGIN_RING(chan, celsius, NV10TCL_VERTEX_COL_4F_R, 4);
OUT_RINGf (chan, 1.0);
--
1.6.4.4
Seemingly Similar Threads
- [PATCH/Gallium] nv50: update nv50_clear to new interface
- [PATCH 1/3] nv50: remove vtxbuf stateobject after a referenced vtxbuf is mapped
- Some cosmetic NV10TCL method changes.
- [PATCH] nv30/exa : cleanup from nv40 exa
- [PATCH 1/4] exa/nv10: use same clip settings as mesa driver