summaryrefslogtreecommitdiff
path: root/src/gallium/auxiliary
diff options
context:
space:
mode:
Diffstat (limited to 'src/gallium/auxiliary')
-rw-r--r--src/gallium/auxiliary/draw/draw_pipe_clip.c38
-rw-r--r--src/gallium/auxiliary/draw/draw_vs_ppc.c6
-rw-r--r--src/gallium/auxiliary/pipebuffer/pb_buffer_fenced.c34
-rw-r--r--src/gallium/auxiliary/tgsi/tgsi_exec.c342
-rw-r--r--src/gallium/auxiliary/tgsi/tgsi_exec.h11
-rw-r--r--src/gallium/auxiliary/tgsi/tgsi_ppc.c3
-rw-r--r--src/gallium/auxiliary/tgsi/tgsi_sanity.c15
-rw-r--r--src/gallium/auxiliary/tgsi/tgsi_sse2.c64
-rw-r--r--src/gallium/auxiliary/util/u_blitter.c195
-rw-r--r--src/gallium/auxiliary/util/u_debug_dump.c6
-rw-r--r--src/gallium/auxiliary/util/u_format.csv22
-rw-r--r--src/gallium/auxiliary/util/u_rect.c10
-rw-r--r--src/gallium/auxiliary/util/u_rect.h2
-rw-r--r--src/gallium/auxiliary/util/u_simple_shaders.c22
-rw-r--r--src/gallium/auxiliary/util/u_tile.c27
15 files changed, 470 insertions, 327 deletions
diff --git a/src/gallium/auxiliary/draw/draw_pipe_clip.c b/src/gallium/auxiliary/draw/draw_pipe_clip.c
index 205cda5eab..51a6115ebf 100644
--- a/src/gallium/auxiliary/draw/draw_pipe_clip.c
+++ b/src/gallium/auxiliary/draw/draw_pipe_clip.c
@@ -55,7 +55,7 @@
-struct clipper {
+struct clip_stage {
struct draw_stage stage; /**< base class */
/* Basically duplicate some of the flatshading logic here:
@@ -70,9 +70,9 @@ struct clipper {
/* This is a bit confusing:
*/
-static INLINE struct clipper *clipper_stage( struct draw_stage *stage )
+static INLINE struct clip_stage *clip_stage( struct draw_stage *stage )
{
- return (struct clipper *)stage;
+ return (struct clip_stage *)stage;
}
@@ -92,11 +92,12 @@ static void interp_attr( float *fdst,
fdst[3] = LINTERP( t, fout[3], fin[3] );
}
+
static void copy_colors( struct draw_stage *stage,
struct vertex_header *dst,
const struct vertex_header *src )
{
- const struct clipper *clipper = clipper_stage(stage);
+ const struct clip_stage *clipper = clip_stage(stage);
uint i;
for (i = 0; i < clipper->num_color_attribs; i++) {
const uint attr = clipper->color_attribs[i];
@@ -108,7 +109,7 @@ static void copy_colors( struct draw_stage *stage,
/* Interpolate between two vertices to produce a third.
*/
-static void interp( const struct clipper *clip,
+static void interp( const struct clip_stage *clip,
struct vertex_header *dst,
float t,
const struct vertex_header *out,
@@ -179,7 +180,7 @@ static void emit_poly( struct draw_stage *stage,
header.v[2] = inlist[0]; /* keep in v[2] for flatshading */
if (i == n-1)
- header.flags |= edge_last;
+ header.flags |= edge_last;
if (0) {
const struct draw_vertex_shader *vs = stage->draw->vs.vertex_shader;
@@ -200,13 +201,14 @@ static void emit_poly( struct draw_stage *stage,
}
}
+
static INLINE float
dot4(const float *a, const float *b)
{
- return (a[0]*b[0] +
- a[1]*b[1] +
- a[2]*b[2] +
- a[3]*b[3]);
+ return (a[0] * b[0] +
+ a[1] * b[1] +
+ a[2] * b[2] +
+ a[3] * b[3]);
}
@@ -217,7 +219,7 @@ do_clip_tri( struct draw_stage *stage,
struct prim_header *header,
unsigned clipmask )
{
- struct clipper *clipper = clipper_stage( stage );
+ struct clip_stage *clipper = clip_stage( stage );
struct vertex_header *a[MAX_CLIPPED_VERTICES];
struct vertex_header *b[MAX_CLIPPED_VERTICES];
struct vertex_header **inlist = a;
@@ -280,6 +282,7 @@ do_clip_tri( struct draw_stage *stage,
dp_prev = dp;
}
+ /* swap in/out lists */
{
struct vertex_header **tmp = inlist;
inlist = outlist;
@@ -291,15 +294,11 @@ do_clip_tri( struct draw_stage *stage,
/* If flat-shading, copy color to new provoking vertex.
*/
if (clipper->flat && inlist[0] != header->v[2]) {
- if (1) {
- inlist[0] = dup_vert(stage, inlist[0], tmpnr++);
- }
+ inlist[0] = dup_vert(stage, inlist[0], tmpnr++);
copy_colors(stage, inlist[0], header->v[2]);
}
-
-
/* Emit the polygon as triangles to the setup stage:
*/
if (n >= 3)
@@ -314,7 +313,7 @@ do_clip_line( struct draw_stage *stage,
struct prim_header *header,
unsigned clipmask )
{
- const struct clipper *clipper = clipper_stage( stage );
+ const struct clip_stage *clipper = clip_stage( stage );
struct vertex_header *v0 = header->v[0];
struct vertex_header *v1 = header->v[1];
const float *pos0 = v0->clip;
@@ -416,13 +415,14 @@ clip_tri( struct draw_stage *stage,
}
}
+
/* Update state. Could further delay this until we hit the first
* primitive that really requires clipping.
*/
static void
clip_init_state( struct draw_stage *stage )
{
- struct clipper *clipper = clipper_stage( stage );
+ struct clip_stage *clipper = clip_stage( stage );
clipper->flat = stage->draw->rasterizer->flatshade ? TRUE : FALSE;
@@ -488,7 +488,7 @@ static void clip_destroy( struct draw_stage *stage )
*/
struct draw_stage *draw_clip_stage( struct draw_context *draw )
{
- struct clipper *clipper = CALLOC_STRUCT(clipper);
+ struct clip_stage *clipper = CALLOC_STRUCT(clip_stage);
if (clipper == NULL)
goto fail;
diff --git a/src/gallium/auxiliary/draw/draw_vs_ppc.c b/src/gallium/auxiliary/draw/draw_vs_ppc.c
index ad184bd696..da9f3e3d35 100644
--- a/src/gallium/auxiliary/draw/draw_vs_ppc.c
+++ b/src/gallium/auxiliary/draw/draw_vs_ppc.c
@@ -98,9 +98,9 @@ vs_ppc_run_linear( struct draw_vertex_shader *base,
/* loop over verts */
for (i = 0; i < count; i += MAX_VERTICES) {
const uint max_vertices = MIN2(MAX_VERTICES, count - i);
- float inputs_soa[PIPE_MAX_SHADER_INPUTS][4][4] ALIGN16_ATTRIB;
- float outputs_soa[PIPE_MAX_SHADER_OUTPUTS][4][4] ALIGN16_ATTRIB;
- float temps_soa[TGSI_EXEC_NUM_TEMPS][4][4] ALIGN16_ATTRIB;
+ PIPE_ALIGN_VAR(16) float inputs_soa[PIPE_MAX_SHADER_INPUTS][4][4];
+ PIPE_ALIGN_VAR(16) float outputs_soa[PIPE_MAX_SHADER_OUTPUTS][4][4];
+ PIPE_ALIGN_VAR(16) float temps_soa[TGSI_EXEC_NUM_TEMPS][4][4];
uint attr;
/* convert (up to) four input verts to SoA format */
diff --git a/src/gallium/auxiliary/pipebuffer/pb_buffer_fenced.c b/src/gallium/auxiliary/pipebuffer/pb_buffer_fenced.c
index a9375abd21..ba6f7b15f9 100644
--- a/src/gallium/auxiliary/pipebuffer/pb_buffer_fenced.c
+++ b/src/gallium/auxiliary/pipebuffer/pb_buffer_fenced.c
@@ -80,7 +80,7 @@ struct fenced_buffer_list
*/
struct fenced_buffer
{
- /*
+ /*
* Immutable members.
*/
@@ -126,8 +126,8 @@ fenced_buffer(struct pb_buffer *buf)
/**
* Add the buffer to the fenced list.
*
- * fenced_buffer_list::mutex and fenced_buffer::mutex must be held, in this
- * order before calling this function.
+ * fenced_buffer_list::mutex and fenced_buffer::mutex must be held, in this
+ * order, before calling this function.
*
* Reference count should be incremented before calling this function.
*/
@@ -191,7 +191,7 @@ fenced_buffer_remove_locked(struct fenced_buffer_list *fenced_list,
* Wait for the fence to expire, and remove it from the fenced list.
*
* fenced_buffer::mutex must be held. fenced_buffer_list::mutex must not be
- * held -- it will
+ * held -- it will be acquired internally.
*/
static INLINE enum pipe_error
fenced_buffer_finish_locked(struct fenced_buffer_list *fenced_list,
@@ -207,7 +207,10 @@ fenced_buffer_finish_locked(struct fenced_buffer_list *fenced_list,
assert(pipe_is_referenced(&fenced_buf->base.base.reference));
assert(fenced_buf->fence);
- /* Acquire the global lock */
+ /*
+ * Acquire the global lock. Must release buffer mutex first to preserve
+ * lock order.
+ */
pipe_mutex_unlock(fenced_buf->mutex);
pipe_mutex_lock(fenced_list->mutex);
pipe_mutex_lock(fenced_buf->mutex);
@@ -217,7 +220,7 @@ fenced_buffer_finish_locked(struct fenced_buffer_list *fenced_list,
/* Remove from the fenced list */
/* TODO: remove consequents */
fenced_buffer_remove_locked(fenced_list, fenced_buf);
-
+
p_atomic_dec(&fenced_buf->base.base.reference.count);
assert(pipe_is_referenced(&fenced_buf->base.base.reference));
@@ -238,7 +241,7 @@ fenced_buffer_finish_locked(struct fenced_buffer_list *fenced_list,
*/
static void
fenced_buffer_list_check_free_locked(struct fenced_buffer_list *fenced_list,
- int wait)
+ int wait)
{
struct pb_fence_ops *ops = fenced_list->ops;
struct list_head *curr, *next;
@@ -274,7 +277,6 @@ fenced_buffer_list_check_free_locked(struct fenced_buffer_list *fenced_list,
pb_buf = &fenced_buf->base;
pb_reference(&pb_buf, NULL);
-
curr = next;
next = curr->next;
@@ -329,7 +331,7 @@ fenced_buffer_map(struct pb_buffer *buf,
if((flags & PIPE_BUFFER_USAGE_DONTBLOCK) &&
ops->fence_signalled(ops, fenced_buf->fence, 0) == 0) {
/* Don't wait for the GPU to finish writing */
- goto finish;
+ goto done;
}
/* Wait for the GPU to finish writing */
@@ -350,7 +352,7 @@ fenced_buffer_map(struct pb_buffer *buf,
fenced_buf->flags |= flags & PIPE_BUFFER_USAGE_CPU_READ_WRITE;
}
-finish:
+done:
pipe_mutex_unlock(fenced_buf->mutex);
return map;
@@ -391,7 +393,7 @@ fenced_buffer_validate(struct pb_buffer *buf,
fenced_buf->vl = NULL;
fenced_buf->validation_flags = 0;
ret = PIPE_OK;
- goto finish;
+ goto done;
}
assert(flags & PIPE_BUFFER_USAGE_GPU_READ_WRITE);
@@ -401,7 +403,7 @@ fenced_buffer_validate(struct pb_buffer *buf,
/* Buffer cannot be validated in two different lists */
if(fenced_buf->vl && fenced_buf->vl != vl) {
ret = PIPE_ERROR_RETRY;
- goto finish;
+ goto done;
}
#if 0
@@ -409,7 +411,7 @@ fenced_buffer_validate(struct pb_buffer *buf,
if(fenced_buf->flags & PIPE_BUFFER_USAGE_CPU_READ_WRITE) {
/* TODO: wait for the thread that mapped the buffer to unmap it */
ret = PIPE_ERROR_RETRY;
- goto finish;
+ goto done;
}
/* Final sanity checking */
assert(!(fenced_buf->flags & PIPE_BUFFER_USAGE_CPU_READ_WRITE));
@@ -420,17 +422,17 @@ fenced_buffer_validate(struct pb_buffer *buf,
(fenced_buf->validation_flags & flags) == flags) {
/* Nothing to do -- buffer already validated */
ret = PIPE_OK;
- goto finish;
+ goto done;
}
ret = pb_validate(fenced_buf->buffer, vl, flags);
if (ret != PIPE_OK)
- goto finish;
+ goto done;
fenced_buf->vl = vl;
fenced_buf->validation_flags |= flags;
-finish:
+done:
pipe_mutex_unlock(fenced_buf->mutex);
return ret;
diff --git a/src/gallium/auxiliary/tgsi/tgsi_exec.c b/src/gallium/auxiliary/tgsi/tgsi_exec.c
index f43233bdb4..2bcb33392a 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_exec.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_exec.c
@@ -1509,7 +1509,8 @@ fetch_texel( struct tgsi_sampler *sampler,
const union tgsi_exec_channel *s,
const union tgsi_exec_channel *t,
const union tgsi_exec_channel *p,
- float lodbias, /* XXX should be float[4] */
+ const union tgsi_exec_channel *c0,
+ enum tgsi_sampler_control control,
union tgsi_exec_channel *r,
union tgsi_exec_channel *g,
union tgsi_exec_channel *b,
@@ -1518,7 +1519,7 @@ fetch_texel( struct tgsi_sampler *sampler,
uint j;
float rgba[NUM_CHANNELS][QUAD_SIZE];
- sampler->get_samples(sampler, s->f, t->f, p->f, lodbias, rgba);
+ sampler->get_samples(sampler, s->f, t->f, p->f, c0->f, control, rgba);
for (j = 0; j < 4; j++) {
r->f[j] = rgba[0][j];
@@ -1529,102 +1530,95 @@ fetch_texel( struct tgsi_sampler *sampler,
}
+#define TEX_MODIFIER_NONE 0
+#define TEX_MODIFIER_PROJECTED 1
+#define TEX_MODIFIER_LOD_BIAS 2
+#define TEX_MODIFIER_EXPLICIT_LOD 3
+
+
static void
exec_tex(struct tgsi_exec_machine *mach,
const struct tgsi_full_instruction *inst,
- boolean biasLod,
- boolean projected)
+ uint modifier)
{
const uint unit = inst->Src[1].Register.Index;
union tgsi_exec_channel r[4];
+ const union tgsi_exec_channel *lod = &ZeroVec;
+ enum tgsi_sampler_control control;
uint chan_index;
- float lodBias;
- /* debug_printf("Sampler %u unit %u\n", sampler, unit); */
+ if (modifier != TEX_MODIFIER_NONE) {
+ FETCH(&r[3], 0, CHAN_W);
+ if (modifier != TEX_MODIFIER_PROJECTED) {
+ lod = &r[3];
+ }
+ }
+
+ if (modifier == TEX_MODIFIER_EXPLICIT_LOD) {
+ control = tgsi_sampler_lod_explicit;
+ } else {
+ control = tgsi_sampler_lod_bias;
+ }
switch (inst->Texture.Texture) {
case TGSI_TEXTURE_1D:
case TGSI_TEXTURE_SHADOW1D:
-
FETCH(&r[0], 0, CHAN_X);
- if (projected) {
- FETCH(&r[1], 0, CHAN_W);
- micro_div( &r[0], &r[0], &r[1] );
+ if (modifier == TEX_MODIFIER_PROJECTED) {
+ micro_div(&r[0], &r[0], &r[3]);
}
- if (biasLod) {
- FETCH(&r[1], 0, CHAN_W);
- lodBias = r[2].f[0];
- }
- else
- lodBias = 0.0;
-
fetch_texel(mach->Samplers[unit],
- &r[0], &ZeroVec, &ZeroVec, lodBias, /* S, T, P, BIAS */
- &r[0], &r[1], &r[2], &r[3]); /* R, G, B, A */
+ &r[0], &ZeroVec, &ZeroVec, lod, /* S, T, P, LOD */
+ control,
+ &r[0], &r[1], &r[2], &r[3]); /* R, G, B, A */
break;
case TGSI_TEXTURE_2D:
case TGSI_TEXTURE_RECT:
case TGSI_TEXTURE_SHADOW2D:
case TGSI_TEXTURE_SHADOWRECT:
-
FETCH(&r[0], 0, CHAN_X);
FETCH(&r[1], 0, CHAN_Y);
FETCH(&r[2], 0, CHAN_Z);
- if (projected) {
- FETCH(&r[3], 0, CHAN_W);
- micro_div( &r[0], &r[0], &r[3] );
- micro_div( &r[1], &r[1], &r[3] );
- micro_div( &r[2], &r[2], &r[3] );
- }
-
- if (biasLod) {
- FETCH(&r[3], 0, CHAN_W);
- lodBias = r[3].f[0];
+ if (modifier == TEX_MODIFIER_PROJECTED) {
+ micro_div(&r[0], &r[0], &r[3]);
+ micro_div(&r[1], &r[1], &r[3]);
+ micro_div(&r[2], &r[2], &r[3]);
}
- else
- lodBias = 0.0;
fetch_texel(mach->Samplers[unit],
- &r[0], &r[1], &r[2], lodBias, /* inputs */
+ &r[0], &r[1], &r[2], lod, /* S, T, P, LOD */
+ control,
&r[0], &r[1], &r[2], &r[3]); /* outputs */
break;
case TGSI_TEXTURE_3D:
case TGSI_TEXTURE_CUBE:
-
FETCH(&r[0], 0, CHAN_X);
FETCH(&r[1], 0, CHAN_Y);
FETCH(&r[2], 0, CHAN_Z);
- if (projected) {
- FETCH(&r[3], 0, CHAN_W);
- micro_div( &r[0], &r[0], &r[3] );
- micro_div( &r[1], &r[1], &r[3] );
- micro_div( &r[2], &r[2], &r[3] );
- }
-
- if (biasLod) {
- FETCH(&r[3], 0, CHAN_W);
- lodBias = r[3].f[0];
+ if (modifier == TEX_MODIFIER_PROJECTED) {
+ micro_div(&r[0], &r[0], &r[3]);
+ micro_div(&r[1], &r[1], &r[3]);
+ micro_div(&r[2], &r[2], &r[3]);
}
- else
- lodBias = 0.0;
fetch_texel(mach->Samplers[unit],
- &r[0], &r[1], &r[2], lodBias,
+ &r[0], &r[1], &r[2], lod,
+ control,
&r[0], &r[1], &r[2], &r[3]);
break;
default:
- assert (0);
+ assert(0);
}
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- STORE( &r[chan_index], 0, chan_index );
+ FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
+ STORE(&r[chan_index], 0, chan_index);
}
}
@@ -1647,8 +1641,9 @@ exec_txd(struct tgsi_exec_machine *mach,
FETCH(&r[0], 0, CHAN_X);
fetch_texel(mach->Samplers[unit],
- &r[0], &ZeroVec, &ZeroVec, 0.0f, /* S, T, P, BIAS */
- &r[0], &r[1], &r[2], &r[3]); /* R, G, B, A */
+ &r[0], &ZeroVec, &ZeroVec, &ZeroVec, /* S, T, P, BIAS */
+ tgsi_sampler_lod_bias,
+ &r[0], &r[1], &r[2], &r[3]); /* R, G, B, A */
break;
case TGSI_TEXTURE_2D:
@@ -1661,8 +1656,9 @@ exec_txd(struct tgsi_exec_machine *mach,
FETCH(&r[2], 0, CHAN_Z);
fetch_texel(mach->Samplers[unit],
- &r[0], &r[1], &r[2], 0.0f, /* inputs */
- &r[0], &r[1], &r[2], &r[3]); /* outputs */
+ &r[0], &r[1], &r[2], &ZeroVec, /* inputs */
+ tgsi_sampler_lod_bias,
+ &r[0], &r[1], &r[2], &r[3]); /* outputs */
break;
case TGSI_TEXTURE_3D:
@@ -1673,7 +1669,8 @@ exec_txd(struct tgsi_exec_machine *mach,
FETCH(&r[2], 0, CHAN_Z);
fetch_texel(mach->Samplers[unit],
- &r[0], &r[1], &r[2], 0.0f,
+ &r[0], &r[1], &r[2], &ZeroVec,
+ tgsi_sampler_lod_bias,
&r[0], &r[1], &r[2], &r[3]);
break;
@@ -1918,6 +1915,130 @@ exec_vector_trinary(struct tgsi_exec_machine *mach,
}
static void
+exec_dp3(struct tgsi_exec_machine *mach,
+ const struct tgsi_full_instruction *inst)
+{
+ unsigned int chan;
+ union tgsi_exec_channel arg[3];
+
+ fetch_source(mach, &arg[0], &inst->Src[0], CHAN_X, TGSI_EXEC_DATA_FLOAT);
+ fetch_source(mach, &arg[1], &inst->Src[1], CHAN_X, TGSI_EXEC_DATA_FLOAT);
+ micro_mul(&arg[2], &arg[0], &arg[1]);
+
+ for (chan = CHAN_Y; chan <= CHAN_Z; chan++) {
+ fetch_source(mach, &arg[0], &inst->Src[0], chan, TGSI_EXEC_DATA_FLOAT);
+ fetch_source(mach, &arg[1], &inst->Src[1], chan, TGSI_EXEC_DATA_FLOAT);
+ micro_mad(&arg[2], arg);
+ }
+
+ for (chan = 0; chan < NUM_CHANNELS; chan++) {
+ if (inst->Dst[0].Register.WriteMask & (1 << chan)) {
+ store_dest(mach, &arg[2], &inst->Dst[0], inst, chan, TGSI_EXEC_DATA_FLOAT);
+ }
+ }
+}
+
+static void
+exec_dp4(struct tgsi_exec_machine *mach,
+ const struct tgsi_full_instruction *inst)
+{
+ unsigned int chan;
+ union tgsi_exec_channel arg[3];
+
+ fetch_source(mach, &arg[0], &inst->Src[0], CHAN_X, TGSI_EXEC_DATA_FLOAT);
+ fetch_source(mach, &arg[1], &inst->Src[1], CHAN_X, TGSI_EXEC_DATA_FLOAT);
+ micro_mul(&arg[2], &arg[0], &arg[1]);
+
+ for (chan = CHAN_Y; chan <= CHAN_W; chan++) {
+ fetch_source(mach, &arg[0], &inst->Src[0], chan, TGSI_EXEC_DATA_FLOAT);
+ fetch_source(mach, &arg[1], &inst->Src[1], chan, TGSI_EXEC_DATA_FLOAT);
+ micro_mad(&arg[2], arg);
+ }
+
+ for (chan = 0; chan < NUM_CHANNELS; chan++) {
+ if (inst->Dst[0].Register.WriteMask & (1 << chan)) {
+ store_dest(mach, &arg[2], &inst->Dst[0], inst, chan, TGSI_EXEC_DATA_FLOAT);
+ }
+ }
+}
+
+static void
+exec_dp2a(struct tgsi_exec_machine *mach,
+ const struct tgsi_full_instruction *inst)
+{
+ unsigned int chan;
+ union tgsi_exec_channel arg[3];
+
+ fetch_source(mach, &arg[0], &inst->Src[0], CHAN_X, TGSI_EXEC_DATA_FLOAT);
+ fetch_source(mach, &arg[1], &inst->Src[1], CHAN_X, TGSI_EXEC_DATA_FLOAT);
+ micro_mul(&arg[2], &arg[0], &arg[1]);
+
+ fetch_source(mach, &arg[0], &inst->Src[0], CHAN_Y, TGSI_EXEC_DATA_FLOAT);
+ fetch_source(mach, &arg[1], &inst->Src[1], CHAN_Y, TGSI_EXEC_DATA_FLOAT);
+ micro_mad(&arg[0], arg);
+
+ fetch_source(mach, &arg[1], &inst->Src[2], CHAN_X, TGSI_EXEC_DATA_FLOAT);
+ micro_add(&arg[0], &arg[0], &arg[1]);
+
+ for (chan = 0; chan < NUM_CHANNELS; chan++) {
+ if (inst->Dst[0].Register.WriteMask & (1 << chan)) {
+ store_dest(mach, &arg[0], &inst->Dst[0], inst, chan, TGSI_EXEC_DATA_FLOAT);
+ }
+ }
+}
+
+static void
+exec_dph(struct tgsi_exec_machine *mach,
+ const struct tgsi_full_instruction *inst)
+{
+ unsigned int chan;
+ union tgsi_exec_channel arg[3];
+
+ fetch_source(mach, &arg[0], &inst->Src[0], CHAN_X, TGSI_EXEC_DATA_FLOAT);
+ fetch_source(mach, &arg[1], &inst->Src[1], CHAN_X, TGSI_EXEC_DATA_FLOAT);
+ micro_mul(&arg[2], &arg[0], &arg[1]);
+
+ fetch_source(mach, &arg[0], &inst->Src[0], CHAN_Y, TGSI_EXEC_DATA_FLOAT);
+ fetch_source(mach, &arg[1], &inst->Src[1], CHAN_Y, TGSI_EXEC_DATA_FLOAT);
+ micro_mad(&arg[2], arg);
+
+ fetch_source(mach, &arg[0], &inst->Src[0], CHAN_Z, TGSI_EXEC_DATA_FLOAT);
+ fetch_source(mach, &arg[1], &inst->Src[1], CHAN_Z, TGSI_EXEC_DATA_FLOAT);
+ micro_mad(&arg[0], arg);
+
+ fetch_source(mach, &arg[1], &inst->Src[1], CHAN_W, TGSI_EXEC_DATA_FLOAT);
+ micro_add(&arg[0], &arg[0], &arg[1]);
+
+ for (chan = 0; chan < NUM_CHANNELS; chan++) {
+ if (inst->Dst[0].Register.WriteMask & (1 << chan)) {
+ store_dest(mach, &arg[0], &inst->Dst[0], inst, chan, TGSI_EXEC_DATA_FLOAT);
+ }
+ }
+}
+
+static void
+exec_dp2(struct tgsi_exec_machine *mach,
+ const struct tgsi_full_instruction *inst)
+{
+ unsigned int chan;
+ union tgsi_exec_channel arg[3];
+
+ fetch_source(mach, &arg[0], &inst->Src[0], CHAN_X, TGSI_EXEC_DATA_FLOAT);
+ fetch_source(mach, &arg[1], &inst->Src[1], CHAN_X, TGSI_EXEC_DATA_FLOAT);
+ micro_mul(&arg[2], &arg[0], &arg[1]);
+
+ fetch_source(mach, &arg[0], &inst->Src[0], CHAN_Y, TGSI_EXEC_DATA_FLOAT);
+ fetch_source(mach, &arg[1], &inst->Src[1], CHAN_Y, TGSI_EXEC_DATA_FLOAT);
+ micro_mad(&arg[2], arg);
+
+ for (chan = 0; chan < NUM_CHANNELS; chan++) {
+ if (inst->Dst[0].Register.WriteMask & (1 << chan)) {
+ store_dest(mach, &arg[2], &inst->Dst[0], inst, chan, TGSI_EXEC_DATA_FLOAT);
+ }
+ }
+}
+
+static void
exec_break(struct tgsi_exec_machine *mach)
{
if (mach->BreakType == TGSI_EXEC_BREAK_INSIDE_LOOP) {
@@ -2397,54 +2518,11 @@ exec_instruction(
break;
case TGSI_OPCODE_DP3:
- /* TGSI_OPCODE_DOT3 */
- FETCH( &r[0], 0, CHAN_X );
- FETCH( &r[1], 1, CHAN_X );
- micro_mul( &r[0], &r[0], &r[1] );
-
- FETCH( &r[1], 0, CHAN_Y );
- FETCH( &r[2], 1, CHAN_Y );
- micro_mul( &r[1], &r[1], &r[2] );
- micro_add( &r[0], &r[0], &r[1] );
-
- FETCH( &r[1], 0, CHAN_Z );
- FETCH( &r[2], 1, CHAN_Z );
- micro_mul( &r[1], &r[1], &r[2] );
- micro_add( &r[0], &r[0], &r[1] );
-
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- STORE( &r[0], 0, chan_index );
- }
+ exec_dp3(mach, inst);
break;
- case TGSI_OPCODE_DP4:
- /* TGSI_OPCODE_DOT4 */
- FETCH(&r[0], 0, CHAN_X);
- FETCH(&r[1], 1, CHAN_X);
-
- micro_mul( &r[0], &r[0], &r[1] );
-
- FETCH(&r[1], 0, CHAN_Y);
- FETCH(&r[2], 1, CHAN_Y);
-
- micro_mul( &r[1], &r[1], &r[2] );
- micro_add( &r[0], &r[0], &r[1] );
-
- FETCH(&r[1], 0, CHAN_Z);
- FETCH(&r[2], 1, CHAN_Z);
-
- micro_mul( &r[1], &r[1], &r[2] );
- micro_add( &r[0], &r[0], &r[1] );
-
- FETCH(&r[1], 0, CHAN_W);
- FETCH(&r[2], 1, CHAN_W);
-
- micro_mul( &r[1], &r[1], &r[2] );
- micro_add( &r[0], &r[0], &r[1] );
-
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- STORE( &r[0], 0, chan_index );
- }
+ case TGSI_OPCODE_DP4:
+ exec_dp4(mach, inst);
break;
case TGSI_OPCODE_DST:
@@ -2540,21 +2618,7 @@ exec_instruction(
break;
case TGSI_OPCODE_DP2A:
- FETCH( &r[0], 0, CHAN_X );
- FETCH( &r[1], 1, CHAN_X );
- micro_mul( &r[0], &r[0], &r[1] );
-
- FETCH( &r[1], 0, CHAN_Y );
- FETCH( &r[2], 1, CHAN_Y );
- micro_mul( &r[1], &r[1], &r[2] );
- micro_add( &r[0], &r[0], &r[1] );
-
- FETCH( &r[2], 2, CHAN_X );
- micro_add( &r[0], &r[0], &r[2] );
-
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- STORE( &r[0], 0, chan_index );
- }
+ exec_dp2a(mach, inst);
break;
case TGSI_OPCODE_FRC:
@@ -2654,30 +2718,7 @@ exec_instruction(
break;
case TGSI_OPCODE_DPH:
- FETCH(&r[0], 0, CHAN_X);
- FETCH(&r[1], 1, CHAN_X);
-
- micro_mul( &r[0], &r[0], &r[1] );
-
- FETCH(&r[1], 0, CHAN_Y);
- FETCH(&r[2], 1, CHAN_Y);
-
- micro_mul( &r[1], &r[1], &r[2] );
- micro_add( &r[0], &r[0], &r[1] );
-
- FETCH(&r[1], 0, CHAN_Z);
- FETCH(&r[2], 1, CHAN_Z);
-
- micro_mul( &r[1], &r[1], &r[2] );
- micro_add( &r[0], &r[0], &r[1] );
-
- FETCH(&r[1], 1, CHAN_W);
-
- micro_add( &r[0], &r[0], &r[1] );
-
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- STORE( &r[0], 0, chan_index );
- }
+ exec_dph(mach, inst);
break;
case TGSI_OPCODE_COS:
@@ -2801,14 +2842,14 @@ exec_instruction(
/* simple texture lookup */
/* src[0] = texcoord */
/* src[1] = sampler unit */
- exec_tex(mach, inst, FALSE, FALSE);
+ exec_tex(mach, inst, TEX_MODIFIER_NONE);
break;
case TGSI_OPCODE_TXB:
/* Texture lookup with lod bias */
/* src[0] = texcoord (src[0].w = LOD bias) */
/* src[1] = sampler unit */
- exec_tex(mach, inst, TRUE, FALSE);
+ exec_tex(mach, inst, TEX_MODIFIER_LOD_BIAS);
break;
case TGSI_OPCODE_TXD:
@@ -2824,14 +2865,14 @@ exec_instruction(
/* Texture lookup with explit LOD */
/* src[0] = texcoord (src[0].w = LOD) */
/* src[1] = sampler unit */
- exec_tex(mach, inst, TRUE, FALSE);
+ exec_tex(mach, inst, TEX_MODIFIER_EXPLICIT_LOD);
break;
case TGSI_OPCODE_TXP:
/* Texture lookup with projection */
/* src[0] = texcoord (src[0].w = projection) */
/* src[1] = sampler unit */
- exec_tex(mach, inst, FALSE, TRUE);
+ exec_tex(mach, inst, TEX_MODIFIER_PROJECTED);
break;
case TGSI_OPCODE_UP2H:
@@ -3091,18 +3132,7 @@ exec_instruction(
break;
case TGSI_OPCODE_DP2:
- FETCH( &r[0], 0, CHAN_X );
- FETCH( &r[1], 1, CHAN_X );
- micro_mul( &r[0], &r[0], &r[1] );
-
- FETCH( &r[1], 0, CHAN_Y );
- FETCH( &r[2], 1, CHAN_Y );
- micro_mul( &r[1], &r[1], &r[2] );
- micro_add( &r[0], &r[0], &r[1] );
-
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- STORE( &r[0], 0, chan_index );
- }
+ exec_dp2(mach, inst);
break;
case TGSI_OPCODE_IF:
diff --git a/src/gallium/auxiliary/tgsi/tgsi_exec.h b/src/gallium/auxiliary/tgsi/tgsi_exec.h
index aa3a98d7f1..59e3b445cc 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_exec.h
+++ b/src/gallium/auxiliary/tgsi/tgsi_exec.h
@@ -2,6 +2,7 @@
*
* Copyright 2007-2008 Tungsten Graphics, Inc., Cedar Park, Texas.
* All Rights Reserved.
+ * Copyright 2009-2010 VMware, Inc. All rights Reserved.
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the
@@ -35,11 +36,13 @@
extern "C" {
#endif
+
#define MAX_LABELS (4 * 1024) /**< basically, max instructions */
#define NUM_CHANNELS 4 /* R,G,B,A */
#define QUAD_SIZE 4 /* 4 pixel/quad */
+
/**
* Registers may be treated as float, signed int or unsigned int.
*/
@@ -69,6 +72,11 @@ struct tgsi_interp_coef
float dady[NUM_CHANNELS];
};
+enum tgsi_sampler_control {
+ tgsi_sampler_lod_bias,
+ tgsi_sampler_lod_explicit
+};
+
/**
* Information for sampling textures, which must be implemented
* by code outside the TGSI executor.
@@ -80,7 +88,8 @@ struct tgsi_sampler
const float s[QUAD_SIZE],
const float t[QUAD_SIZE],
const float p[QUAD_SIZE],
- float lodbias,
+ const float c0[QUAD_SIZE],
+ enum tgsi_sampler_control control,
float rgba[NUM_CHANNELS][QUAD_SIZE]);
};
diff --git a/src/gallium/auxiliary/tgsi/tgsi_ppc.c b/src/gallium/auxiliary/tgsi/tgsi_ppc.c
index 138d2d095b..ad553c71a5 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_ppc.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_ppc.c
@@ -51,7 +51,8 @@
* Since it's pretty much impossible to form PPC vector immediates, load
* them from memory here:
*/
-const float ppc_builtin_constants[] ALIGN16_ATTRIB = {
+PIPE_ALIGN_VAR(16) const float
+ppc_builtin_constants[] = {
1.0f, -128.0f, 128.0, 0.0
};
diff --git a/src/gallium/auxiliary/tgsi/tgsi_sanity.c b/src/gallium/auxiliary/tgsi/tgsi_sanity.c
index 9b0644465a..7f1c8e5dd6 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_sanity.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_sanity.c
@@ -265,13 +265,15 @@ check_register_usage(
}
else {
if (!is_register_declared( ctx, reg )) {
- if (reg->dimensions == 2)
+ if (reg->dimensions == 2) {
report_error( ctx, "%s[%d][%d]: Undeclared %s register", file_names[reg->file],
reg->indices[0], reg->indices[1], name );
- else
+ }
+ else {
report_error( ctx, "%s[%d]: Undeclared %s register", file_names[reg->file],
reg->indices[0], name );
}
+ }
if (!is_register_used( ctx, reg ))
cso_hash_insert(ctx->regs_used, scan_register_key(reg), reg);
else
@@ -333,15 +335,15 @@ iter_instruction(
fill_scan_register1d(ind_reg,
inst->Src[i].Indirect.File,
inst->Src[i].Indirect.Index);
+ if (!(reg->file == TGSI_FILE_ADDRESS || reg->file == TGSI_FILE_LOOP) ||
+ reg->indices[0] != 0) {
+ report_warning(ctx, "Indirect register neither ADDR[0] nor LOOP[0]");
+ }
check_register_usage(
ctx,
reg,
"indirect",
FALSE );
- if (!(reg->file == TGSI_FILE_ADDRESS || reg->file == TGSI_FILE_LOOP) ||
- reg->indices[0] != 0) {
- report_warning(ctx, "Indirect register neither ADDR[0] nor LOOP[0]");
- }
}
}
@@ -513,6 +515,7 @@ regs_hash_destroy(struct cso_hash *hash)
while (!cso_hash_iter_is_null(iter)) {
scan_register *reg = (scan_register *)cso_hash_iter_data(iter);
iter = cso_hash_erase(hash, iter);
+ assert(reg->file < TGSI_FILE_COUNT);
FREE(reg);
}
cso_hash_delete(hash);
diff --git a/src/gallium/auxiliary/tgsi/tgsi_sse2.c b/src/gallium/auxiliary/tgsi/tgsi_sse2.c
index 118059ace9..a85cc4659e 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_sse2.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_sse2.c
@@ -2,6 +2,7 @@
*
* Copyright 2007-2008 Tungsten Graphics, Inc., Cedar Park, Texas.
* All Rights Reserved.
+ * Copyright 2009-2010 VMware, Inc. All rights Reserved.
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the
@@ -1418,13 +1419,13 @@ fetch_texel( struct tgsi_sampler **sampler,
sampler, *sampler,
store );
- debug_printf("lodbias %f\n", store[12]);
-
for (j = 0; j < 4; j++)
- debug_printf("sample %d texcoord %f %f\n",
+ debug_printf("sample %d texcoord %f %f %f lodbias %f\n",
j,
store[0+j],
- store[4+j]);
+ store[4+j],
+ store[8 + j],
+ store[12 + j]);
#endif
{
@@ -1433,7 +1434,8 @@ fetch_texel( struct tgsi_sampler **sampler,
&store[0], /* s */
&store[4], /* t */
&store[8], /* r */
- store[12], /* lodbias */
+ &store[12], /* lodbias */
+ tgsi_sampler_lod_bias,
rgba); /* results */
memcpy( store, rgba, 16 * sizeof(float));
@@ -2144,40 +2146,50 @@ emit_instruction(
break;
case TGSI_OPCODE_XPD:
+ /* Note: we do all stores after all operands have been fetched
+ * to avoid src/dst register aliasing issues for an instruction
+ * such as: XPD TEMP[2].xyz, TEMP[0], TEMP[2];
+ */
if( IS_DST0_CHANNEL_ENABLED( *inst, CHAN_X ) ||
IS_DST0_CHANNEL_ENABLED( *inst, CHAN_Y ) ) {
- FETCH( func, *inst, 1, 1, CHAN_Z );
- FETCH( func, *inst, 3, 0, CHAN_Z );
+ FETCH( func, *inst, 1, 1, CHAN_Z ); /* xmm[1] = src[1].z */
+ FETCH( func, *inst, 3, 0, CHAN_Z ); /* xmm[3] = src[0].z */
}
if( IS_DST0_CHANNEL_ENABLED( *inst, CHAN_X ) ||
IS_DST0_CHANNEL_ENABLED( *inst, CHAN_Z ) ) {
- FETCH( func, *inst, 0, 0, CHAN_Y );
- FETCH( func, *inst, 4, 1, CHAN_Y );
+ FETCH( func, *inst, 0, 0, CHAN_Y ); /* xmm[0] = src[0].y */
+ FETCH( func, *inst, 4, 1, CHAN_Y ); /* xmm[4] = src[1].y */
}
IF_IS_DST0_CHANNEL_ENABLED( *inst, CHAN_X ) {
- emit_MOV( func, 2, 0 );
- emit_mul( func, 2, 1 );
- emit_MOV( func, 5, 3 );
- emit_mul( func, 5, 4 );
- emit_sub( func, 2, 5 );
- STORE( func, *inst, 2, 0, CHAN_X );
+ emit_MOV( func, 7, 0 ); /* xmm[7] = xmm[0] */
+ emit_mul( func, 7, 1 ); /* xmm[7] = xmm[2] * xmm[1] */
+ emit_MOV( func, 5, 3 ); /* xmm[5] = xmm[3] */
+ emit_mul( func, 5, 4 ); /* xmm[5] = xmm[5] * xmm[4] */
+ emit_sub( func, 7, 5 ); /* xmm[7] = xmm[2] - xmm[5] */
+ /* store xmm[7] in dst.x below */
}
if( IS_DST0_CHANNEL_ENABLED( *inst, CHAN_Y ) ||
IS_DST0_CHANNEL_ENABLED( *inst, CHAN_Z ) ) {
- FETCH( func, *inst, 2, 1, CHAN_X );
- FETCH( func, *inst, 5, 0, CHAN_X );
+ FETCH( func, *inst, 2, 1, CHAN_X ); /* xmm[2] = src[1].x */
+ FETCH( func, *inst, 5, 0, CHAN_X ); /* xmm[5] = src[0].x */
}
IF_IS_DST0_CHANNEL_ENABLED( *inst, CHAN_Y ) {
- emit_mul( func, 3, 2 );
- emit_mul( func, 1, 5 );
- emit_sub( func, 3, 1 );
- STORE( func, *inst, 3, 0, CHAN_Y );
+ emit_mul( func, 3, 2 ); /* xmm[3] = xmm[3] * xmm[2] */
+ emit_mul( func, 1, 5 ); /* xmm[1] = xmm[1] * xmm[5] */
+ emit_sub( func, 3, 1 ); /* xmm[3] = xmm[3] - xmm[1] */
+ /* store xmm[3] in dst.y below */
}
IF_IS_DST0_CHANNEL_ENABLED( *inst, CHAN_Z ) {
- emit_mul( func, 5, 4 );
- emit_mul( func, 0, 2 );
- emit_sub( func, 5, 0 );
- STORE( func, *inst, 5, 0, CHAN_Z );
+ emit_mul( func, 5, 4 ); /* xmm[5] = xmm[5] * xmm[4] */
+ emit_mul( func, 0, 2 ); /* xmm[0] = xmm[0] * xmm[2] */
+ emit_sub( func, 5, 0 ); /* xmm[5] = xmm[5] - xmm[0] */
+ STORE( func, *inst, 5, 0, CHAN_Z ); /* dst.z = xmm[5] */
+ }
+ IF_IS_DST0_CHANNEL_ENABLED( *inst, CHAN_X ) {
+ STORE( func, *inst, 7, 0, CHAN_X ); /* dst.x = xmm[7] */
+ }
+ IF_IS_DST0_CHANNEL_ENABLED( *inst, CHAN_Y ) {
+ STORE( func, *inst, 3, 0, CHAN_Y ); /* dst.y = xmm[3] */
}
IF_IS_DST0_CHANNEL_ENABLED( *inst, CHAN_W ) {
emit_tempf(
@@ -2506,7 +2518,7 @@ emit_instruction(
break;
case TGSI_OPCODE_TXL:
- emit_tex( func, inst, TRUE, FALSE );
+ return 0;
break;
case TGSI_OPCODE_TXP:
diff --git a/src/gallium/auxiliary/util/u_blitter.c b/src/gallium/auxiliary/util/u_blitter.c
index 1f794d39a1..249a0375fc 100644
--- a/src/gallium/auxiliary/util/u_blitter.c
+++ b/src/gallium/auxiliary/util/u_blitter.c
@@ -48,6 +48,8 @@
#include "util/u_simple_shaders.h"
#include "util/u_texture.h"
+#define INVALID_PTR ((void*)~0)
+
struct blitter_context_priv
{
struct blitter_context blitter;
@@ -110,6 +112,11 @@ struct blitter_context *util_blitter_create(struct pipe_context *pipe)
ctx->pipe = pipe;
/* init state objects for them to be considered invalid */
+ ctx->blitter.saved_blend_state = INVALID_PTR;
+ ctx->blitter.saved_dsa_state = INVALID_PTR;
+ ctx->blitter.saved_rs_state = INVALID_PTR;
+ ctx->blitter.saved_fs = INVALID_PTR;
+ ctx->blitter.saved_vs = INVALID_PTR;
ctx->blitter.saved_fb_state.nr_cbufs = ~0;
ctx->blitter.saved_num_textures = ~0;
ctx->blitter.saved_num_sampler_states = ~0;
@@ -156,6 +163,7 @@ struct blitter_context *util_blitter_create(struct pipe_context *pipe)
rs_state.cull_mode = PIPE_WINDING_NONE;
rs_state.bypass_vs_clip_and_viewport = 1;
rs_state.gl_rasterization_rules = 1;
+ rs_state.flatshade = 1;
ctx->rs_state = pipe->create_rasterizer_state(pipe, &rs_state);
/* fragment shaders are created on-demand */
@@ -234,11 +242,11 @@ void util_blitter_destroy(struct blitter_context *blitter)
static void blitter_check_saved_CSOs(struct blitter_context_priv *ctx)
{
/* make sure these CSOs have been saved */
- assert(ctx->blitter.saved_blend_state &&
- ctx->blitter.saved_dsa_state &&
- ctx->blitter.saved_rs_state &&
- ctx->blitter.saved_fs &&
- ctx->blitter.saved_vs);
+ assert(ctx->blitter.saved_blend_state != INVALID_PTR &&
+ ctx->blitter.saved_dsa_state != INVALID_PTR &&
+ ctx->blitter.saved_rs_state != INVALID_PTR &&
+ ctx->blitter.saved_fs != INVALID_PTR &&
+ ctx->blitter.saved_vs != INVALID_PTR);
}
static void blitter_restore_CSOs(struct blitter_context_priv *ctx)
@@ -252,11 +260,11 @@ static void blitter_restore_CSOs(struct blitter_context_priv *ctx)
pipe->bind_fs_state(pipe, ctx->blitter.saved_fs);
pipe->bind_vs_state(pipe, ctx->blitter.saved_vs);
- ctx->blitter.saved_blend_state = 0;
- ctx->blitter.saved_dsa_state = 0;
- ctx->blitter.saved_rs_state = 0;
- ctx->blitter.saved_fs = 0;
- ctx->blitter.saved_vs = 0;
+ ctx->blitter.saved_blend_state = INVALID_PTR;
+ ctx->blitter.saved_dsa_state = INVALID_PTR;
+ ctx->blitter.saved_rs_state = INVALID_PTR;
+ ctx->blitter.saved_fs = INVALID_PTR;
+ ctx->blitter.saved_vs = INVALID_PTR;
/* restore the state objects which are required to be saved before copy/fill
*/
@@ -371,9 +379,16 @@ static void blitter_set_texcoords_cube(struct blitter_context_priv *ctx,
float t1 = y1 / (float)surf->height;
float s2 = x2 / (float)surf->width;
float t2 = y2 / (float)surf->height;
- const float st[4][2] = {
- {s1, t1}, {s2, t1}, {s2, t2}, {s1, t2}
- };
+ float st[4][2];
+
+ st[0][0] = s1;
+ st[0][1] = t1;
+ st[1][0] = s2;
+ st[1][1] = t1;
+ st[2][0] = s2;
+ st[2][1] = t2;
+ st[3][0] = s1;
+ st[3][1] = t2;
util_map_texcoords2d_onto_cubemap(surf->face,
/* pointer, stride in floats */
@@ -560,45 +575,29 @@ void util_blitter_clear(struct blitter_context *blitter,
blitter_restore_CSOs(ctx);
}
-void util_blitter_copy(struct blitter_context *blitter,
- struct pipe_surface *dst,
- unsigned dstx, unsigned dsty,
- struct pipe_surface *src,
- unsigned srcx, unsigned srcy,
- unsigned width, unsigned height,
- boolean ignore_stencil)
+static boolean
+is_overlap(unsigned sx1, unsigned sx2, unsigned sy1, unsigned sy2,
+ unsigned dx1, unsigned dx2, unsigned dy1, unsigned dy2)
+{
+ if (sx1 >= dx2 || sx2 <= dx1 || sy1 >= dy2 || sy2 <= dy1) {
+ return FALSE;
+ } else {
+ return TRUE;
+ }
+}
+
+static void util_blitter_do_copy(struct blitter_context *blitter,
+ struct pipe_surface *dst,
+ unsigned dstx, unsigned dsty,
+ struct pipe_surface *src,
+ unsigned srcx, unsigned srcy,
+ unsigned width, unsigned height,
+ boolean is_depth)
{
struct blitter_context_priv *ctx = (struct blitter_context_priv*)blitter;
struct pipe_context *pipe = ctx->pipe;
- struct pipe_screen *screen = pipe->screen;
struct pipe_framebuffer_state fb_state;
- boolean is_stencil, is_depth;
- unsigned dst_tex_usage;
-
- /* give up if textures are not set */
- assert(dst->texture && src->texture);
- if (!dst->texture || !src->texture)
- return;
-
- is_depth = util_format_get_component_bits(src->format, UTIL_FORMAT_COLORSPACE_ZS, 0) != 0;
- is_stencil = util_format_get_component_bits(src->format, UTIL_FORMAT_COLORSPACE_ZS, 1) != 0;
- dst_tex_usage = is_depth || is_stencil ? PIPE_TEXTURE_USAGE_DEPTH_STENCIL :
- PIPE_TEXTURE_USAGE_RENDER_TARGET;
- /* check if we can sample from and render to the surfaces */
- /* (assuming copying a stencil buffer is not possible) */
- if ((!ignore_stencil && is_stencil) ||
- !screen->is_format_supported(screen, dst->format, dst->texture->target,
- dst_tex_usage, 0) ||
- !screen->is_format_supported(screen, src->format, src->texture->target,
- PIPE_TEXTURE_USAGE_SAMPLER, 0)) {
- util_surface_copy(pipe, FALSE, dst, dstx, dsty, src, srcx, srcy,
- width, height);
- return;
- }
-
- /* check whether the states are properly saved */
- blitter_check_saved_CSOs(ctx);
assert(blitter->saved_fb_state.nr_cbufs != ~0);
assert(blitter->saved_num_textures != ~0);
assert(blitter->saved_num_sampler_states != ~0);
@@ -656,6 +655,108 @@ void util_blitter_copy(struct blitter_context *blitter,
blitter_set_rectangle(ctx, dstx, dsty, dstx+width, dsty+height, 0);
blitter_draw_quad(ctx);
+
+}
+
+static void util_blitter_overlap_copy(struct blitter_context *blitter,
+ struct pipe_surface *dst,
+ unsigned dstx, unsigned dsty,
+ struct pipe_surface *src,
+ unsigned srcx, unsigned srcy,
+ unsigned width, unsigned height)
+{
+ struct blitter_context_priv *ctx = (struct blitter_context_priv*)blitter;
+ struct pipe_context *pipe = ctx->pipe;
+ struct pipe_screen *screen = pipe->screen;
+
+ struct pipe_texture texTemp;
+ struct pipe_texture *texture;
+ struct pipe_surface *tex_surf;
+
+ /* check whether the states are properly saved */
+ blitter_check_saved_CSOs(ctx);
+
+ memset(&texTemp, 0, sizeof(texTemp));
+ texTemp.target = PIPE_TEXTURE_2D;
+ texTemp.format = dst->texture->format; /* XXX verify supported by driver! */
+ texTemp.last_level = 0;
+ texTemp.width0 = width;
+ texTemp.height0 = height;
+ texTemp.depth0 = 1;
+
+ texture = screen->texture_create(screen, &texTemp);
+ if (!texture)
+ return;
+
+ tex_surf = screen->get_tex_surface(screen, texture, 0, 0, 0,
+ PIPE_BUFFER_USAGE_GPU_READ |
+ PIPE_BUFFER_USAGE_GPU_WRITE);
+
+ /* blit from the src to the temp */
+ util_blitter_do_copy(blitter, tex_surf, 0, 0,
+ src, srcx, srcy,
+ width, height,
+ FALSE);
+ util_blitter_do_copy(blitter, dst, dstx, dsty,
+ tex_surf, 0, 0,
+ width, height,
+ FALSE);
+ pipe_surface_reference(&tex_surf, NULL);
+ pipe_texture_reference(&texture, NULL);
+ blitter_restore_CSOs(ctx);
+}
+
+void util_blitter_copy(struct blitter_context *blitter,
+ struct pipe_surface *dst,
+ unsigned dstx, unsigned dsty,
+ struct pipe_surface *src,
+ unsigned srcx, unsigned srcy,
+ unsigned width, unsigned height,
+ boolean ignore_stencil)
+{
+ struct blitter_context_priv *ctx = (struct blitter_context_priv*)blitter;
+ struct pipe_context *pipe = ctx->pipe;
+ struct pipe_screen *screen = pipe->screen;
+ boolean is_stencil, is_depth;
+ unsigned dst_tex_usage;
+
+ /* give up if textures are not set */
+ assert(dst->texture && src->texture);
+ if (!dst->texture || !src->texture)
+ return;
+
+ if (dst->texture == src->texture) {
+ if (is_overlap(srcx, srcx + width, srcy, srcy + height,
+ dstx, dstx + width, dsty, dsty + height)) {
+ util_blitter_overlap_copy(blitter, dst, dstx, dsty, src, srcx, srcy,
+ width, height);
+ return;
+ }
+ }
+
+ is_depth = util_format_get_component_bits(src->format, UTIL_FORMAT_COLORSPACE_ZS, 0) != 0;
+ is_stencil = util_format_get_component_bits(src->format, UTIL_FORMAT_COLORSPACE_ZS, 1) != 0;
+ dst_tex_usage = is_depth || is_stencil ? PIPE_TEXTURE_USAGE_DEPTH_STENCIL :
+ PIPE_TEXTURE_USAGE_RENDER_TARGET;
+
+ /* check if we can sample from and render to the surfaces */
+ /* (assuming copying a stencil buffer is not possible) */
+ if ((!ignore_stencil && is_stencil) ||
+ !screen->is_format_supported(screen, dst->format, dst->texture->target,
+ dst_tex_usage, 0) ||
+ !screen->is_format_supported(screen, src->format, src->texture->target,
+ PIPE_TEXTURE_USAGE_SAMPLER, 0)) {
+ util_surface_copy(pipe, FALSE, dst, dstx, dsty, src, srcx, srcy,
+ width, height);
+ return;
+ }
+
+ /* check whether the states are properly saved */
+ blitter_check_saved_CSOs(ctx);
+ util_blitter_do_copy(blitter,
+ dst, dstx, dsty,
+ src, srcx, srcy,
+ width, height, is_depth);
blitter_restore_CSOs(ctx);
}
diff --git a/src/gallium/auxiliary/util/u_debug_dump.c b/src/gallium/auxiliary/util/u_debug_dump.c
index 09866880ae..61624d05c0 100644
--- a/src/gallium/auxiliary/util/u_debug_dump.c
+++ b/src/gallium/auxiliary/util/u_debug_dump.c
@@ -255,15 +255,13 @@ DEFINE_DEBUG_DUMP_CONTINUOUS(tex_mipfilter)
static const char *
debug_dump_tex_filter_names[] = {
"PIPE_TEX_FILTER_NEAREST",
- "PIPE_TEX_FILTER_LINEAR",
- "PIPE_TEX_FILTER_ANISO"
+ "PIPE_TEX_FILTER_LINEAR"
};
static const char *
debug_dump_tex_filter_short_names[] = {
"nearest",
- "linear",
- "aniso"
+ "linear"
};
DEFINE_DEBUG_DUMP_CONTINUOUS(tex_filter)
diff --git a/src/gallium/auxiliary/util/u_format.csv b/src/gallium/auxiliary/util/u_format.csv
index 866b18ff16..9f16b42944 100644
--- a/src/gallium/auxiliary/util/u_format.csv
+++ b/src/gallium/auxiliary/util/u_format.csv
@@ -76,9 +76,9 @@ PIPE_FORMAT_R8G8_SNORM , array , 1, 1, sn8 , sn8 , , , xy01,
PIPE_FORMAT_R8G8B8_SNORM , array , 1, 1, sn8 , sn8 , sn8 , , xyz1, rgb
PIPE_FORMAT_R8G8B8A8_SNORM , array , 1, 1, sn8 , sn8 , sn8 , sn8 , xyzw, rgb
PIPE_FORMAT_R8G8B8X8_SNORM , array , 1, 1, sn8 , sn8 , sn8 , sn8 , xyz1, rgb
-PIPE_FORMAT_B6G5R5_SNORM , arith , 1, 1, sn5 , sn5 , sn6 , , zyx1, rgb
-PIPE_FORMAT_A8B8G8R8_SNORM , arith , 1, 1, sn8 , sn8 , sn8 , sn8 , zyxw, rgb
-PIPE_FORMAT_X8B8G8R8_SNORM , arith , 1, 1, sn8 , sn8 , sn8 , sn8 , zyx1, rgb
+PIPE_FORMAT_B6G5R5_SNORM , arith , 1, 1, sn5 , sn5 , sn6 , , xyz1, rgb
+PIPE_FORMAT_A8B8G8R8_SNORM , array , 1, 1, sn8 , sn8 , sn8 , sn8 , wzyx, rgb
+PIPE_FORMAT_X8B8G8R8_SNORM , array , 1, 1, sn8 , sn8 , sn8 , sn8 , wzy1, rgb
PIPE_FORMAT_R8_SSCALED , array , 1, 1, s8 , , , , x001, rgb
PIPE_FORMAT_R8G8_SSCALED , array , 1, 1, s8 , s8 , , , xy01, rgb
PIPE_FORMAT_R8G8B8_SSCALED , array , 1, 1, s8 , s8 , s8 , , xyz1, rgb
@@ -90,14 +90,14 @@ PIPE_FORMAT_R32G32B32_FIXED , array , 1, 1, h32 , h32 , h32 , , xyz1,
PIPE_FORMAT_R32G32B32A32_FIXED , array , 1, 1, h32 , h32 , h32 , h32 , xyzw, rgb
PIPE_FORMAT_L8_SRGB , arith , 1, 1, u8 , , , , xxx1, srgb
PIPE_FORMAT_A8L8_SRGB , arith , 1, 1, u8 , u8 , , , xxxy, srgb
-PIPE_FORMAT_R8G8B8_SRGB , arith , 1, 1, u8 , u8 , u8 , , xyz1, srgb
-PIPE_FORMAT_R8G8B8A8_SRGB , arith , 1, 1, u8 , u8 , u8 , u8 , xyzw, srgb
-PIPE_FORMAT_R8G8B8X8_SRGB , arith , 1, 1, u8 , u8 , u8 , u8 , xyz1, srgb
-PIPE_FORMAT_A8R8G8B8_SRGB , arith , 1, 1, u8 , u8 , u8 , u8 , wxyz, srgb
-PIPE_FORMAT_X8R8G8B8_SRGB , arith , 1, 1, u8 , u8 , u8 , u8 , 1xyz, srgb
-PIPE_FORMAT_B8G8R8A8_SRGB , arith , 1, 1, u8 , u8 , u8 , u8 , zyxw, srgb
-PIPE_FORMAT_B8G8R8X8_SRGB , arith , 1, 1, u8 , u8 , u8 , u8 , zyx1, srgb
-PIPE_FORMAT_X8UB8UG8SR8S_NORM , arith , 1, 1, sn8 , sn8 , un8 , x8 , 1zyx, rgb
+PIPE_FORMAT_R8G8B8_SRGB , array , 1, 1, u8 , u8 , u8 , , xyz1, srgb
+PIPE_FORMAT_R8G8B8A8_SRGB , array , 1, 1, u8 , u8 , u8 , u8 , xyzw, srgb
+PIPE_FORMAT_R8G8B8X8_SRGB , array , 1, 1, u8 , u8 , u8 , u8 , xyz1, srgb
+PIPE_FORMAT_A8R8G8B8_SRGB , array , 1, 1, u8 , u8 , u8 , u8 , yzwx, srgb
+PIPE_FORMAT_X8R8G8B8_SRGB , array , 1, 1, u8 , u8 , u8 , u8 , yzw1, srgb
+PIPE_FORMAT_B8G8R8A8_SRGB , array , 1, 1, u8 , u8 , u8 , u8 , zyxw, srgb
+PIPE_FORMAT_B8G8R8X8_SRGB , array , 1, 1, u8 , u8 , u8 , u8 , zyx1, srgb
+PIPE_FORMAT_X8UB8UG8SR8S_NORM , array , 1, 1, sn8 , sn8 , un8 , x8 , wzy1, rgb
PIPE_FORMAT_B6UG5SR5S_NORM , arith , 1, 1, sn5 , sn5 , un6 , , xyz1, rgb
PIPE_FORMAT_DXT1_RGB , dxt , 4, 4, x64 , , , , xyz1, rgb
PIPE_FORMAT_DXT1_RGBA , dxt , 4, 4, x64 , , , , xyzw, rgb
diff --git a/src/gallium/auxiliary/util/u_rect.c b/src/gallium/auxiliary/util/u_rect.c
index 298fbacecb..8479161c74 100644
--- a/src/gallium/auxiliary/util/u_rect.c
+++ b/src/gallium/auxiliary/util/u_rect.c
@@ -41,7 +41,7 @@
/**
* Copy 2D rect from one place to another.
* Position and sizes are in pixels.
- * src_pitch may be negative to do vertical flip of pixels from source.
+ * src_stride may be negative to do vertical flip of pixels from source.
*/
void
util_copy_rect(ubyte * dst,
@@ -54,7 +54,7 @@ util_copy_rect(ubyte * dst,
const ubyte * src,
int src_stride,
unsigned src_x,
- int src_y)
+ unsigned src_y)
{
unsigned i;
int src_stride_pos = src_stride < 0 ? -src_stride : src_stride;
@@ -65,10 +65,6 @@ util_copy_rect(ubyte * dst,
assert(blocksize > 0);
assert(blockwidth > 0);
assert(blockheight > 0);
- assert(src_x >= 0);
- assert(src_y >= 0);
- assert(dst_x >= 0);
- assert(dst_y >= 0);
dst_x /= blockwidth;
dst_y /= blockheight;
@@ -113,8 +109,6 @@ util_fill_rect(ubyte * dst,
assert(blocksize > 0);
assert(blockwidth > 0);
assert(blockheight > 0);
- assert(dst_x >= 0);
- assert(dst_y >= 0);
dst_x /= blockwidth;
dst_y /= blockheight;
diff --git a/src/gallium/auxiliary/util/u_rect.h b/src/gallium/auxiliary/util/u_rect.h
index 5e444ffae2..b44d821904 100644
--- a/src/gallium/auxiliary/util/u_rect.h
+++ b/src/gallium/auxiliary/util/u_rect.h
@@ -45,7 +45,7 @@ extern void
util_copy_rect(ubyte * dst, enum pipe_format format,
unsigned dst_stride, unsigned dst_x, unsigned dst_y,
unsigned width, unsigned height, const ubyte * src,
- int src_stride, unsigned src_x, int src_y);
+ int src_stride, unsigned src_x, unsigned src_y);
extern void
util_fill_rect(ubyte * dst, enum pipe_format format,
diff --git a/src/gallium/auxiliary/util/u_simple_shaders.c b/src/gallium/auxiliary/util/u_simple_shaders.c
index 8172ead020..b751e29ab6 100644
--- a/src/gallium/auxiliary/util/u_simple_shaders.c
+++ b/src/gallium/auxiliary/util/u_simple_shaders.c
@@ -44,13 +44,15 @@
/**
* Make simple vertex pass-through shader.
+ * \param num_attribs number of attributes to pass through
+ * \param semantic_names array of semantic names for each attribute
+ * \param semantic_indexes array of semantic indexes for each attribute
*/
void *
util_make_vertex_passthrough_shader(struct pipe_context *pipe,
uint num_attribs,
const uint *semantic_names,
const uint *semantic_indexes)
-
{
struct ureg_program *ureg;
uint i;
@@ -78,8 +80,6 @@ util_make_vertex_passthrough_shader(struct pipe_context *pipe,
}
-
-
/**
* Make simple fragment texture shader:
* IMM {0,0,0,1} // (if writemask != 0xf)
@@ -125,6 +125,12 @@ util_make_fragment_tex_shader_writemask(struct pipe_context *pipe,
return ureg_create_shader_and_destroy( ureg, pipe );
}
+
+/**
+ * Make a simple fragment shader that sets the output color to a color
+ * taken from a texture.
+ * \param tex_target one of PIPE_TEXTURE_x
+ */
void *
util_make_fragment_tex_shader(struct pipe_context *pipe, unsigned tex_target )
{
@@ -133,6 +139,7 @@ util_make_fragment_tex_shader(struct pipe_context *pipe, unsigned tex_target )
TGSI_WRITEMASK_XYZW );
}
+
/**
* Make a simple fragment texture shader which reads an X component from
* a texture and writes it as depth.
@@ -177,6 +184,7 @@ util_make_fragment_tex_shader_writedepth(struct pipe_context *pipe,
return ureg_create_shader_and_destroy( ureg, pipe );
}
+
/**
* Make simple fragment color pass-through shader.
*/
@@ -186,15 +194,19 @@ util_make_fragment_passthrough_shader(struct pipe_context *pipe)
return util_make_fragment_clonecolor_shader(pipe, 1);
}
+
+/**
+ * Make a fragment shader that copies the input color to N output colors.
+ */
void *
util_make_fragment_clonecolor_shader(struct pipe_context *pipe, int num_cbufs)
{
struct ureg_program *ureg;
struct ureg_src src;
- struct ureg_dst dst[8];
+ struct ureg_dst dst[PIPE_MAX_COLOR_BUFS];
int i;
- assert(num_cbufs <= 8);
+ assert(num_cbufs <= PIPE_MAX_COLOR_BUFS);
ureg = ureg_create( TGSI_PROCESSOR_FRAGMENT );
if (ureg == NULL)
diff --git a/src/gallium/auxiliary/util/u_tile.c b/src/gallium/auxiliary/util/u_tile.c
index 5b8dd1abb9..1ba82bb21f 100644
--- a/src/gallium/auxiliary/util/u_tile.c
+++ b/src/gallium/auxiliary/util/u_tile.c
@@ -1155,27 +1155,6 @@ ycbcr_get_tile_rgba(const ushort *src,
}
-static void
-fake_get_tile_rgba(const ushort *src,
- unsigned w, unsigned h,
- float *p,
- unsigned dst_stride)
-{
- unsigned i, j;
-
- for (i = 0; i < h; i++) {
- float *pRow = p;
- for (j = 0; j < w; j++, pRow += 4) {
- pRow[0] =
- pRow[1] =
- pRow[2] =
- pRow[3] = (i ^ j) & 1 ? 1.0f : 0.0f;
- }
- p += dst_stride;
- }
-}
-
-
void
pipe_tile_raw_to_rgba(enum pipe_format format,
void *src,
@@ -1258,8 +1237,10 @@ pipe_tile_raw_to_rgba(enum pipe_format format,
ycbcr_get_tile_rgba((ushort *) src, w, h, dst, dst_stride, TRUE);
break;
default:
- debug_printf("%s: unsupported format %s\n", __FUNCTION__, pf_name(format));
- fake_get_tile_rgba(src, w, h, dst, dst_stride);
+ util_format_read_4f(format,
+ dst, dst_stride * sizeof(float),
+ src, util_format_get_stride(format, w),
+ 0, 0, w, h);
}
}
generated by cgit v1.2.3 (git 2.39.1) at 2025-01-26 04:36:19 +0000