KolibriOS/kolibrios
13
4
Fork 12
Code Issues 78 Pull Requests 5 Projects Releases Activity

intel-2D: Ironlake SandyBridge IvyBridge textured blitter

Browse Source 
git-svn-id: svn://kolibrios.org@3280 a494cfbc-eb01-0410-851d-a64ba20cac60
This commit is contained in:
Sergey Semyonov (Serge) 2013-02-27 07:07:21 +00:00
parent 25f867273b
commit 51563dce61
11 changed files with 7633 additions and 13 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

22
drivers/video/Intel-2D/gen4_source.h Normal file
View File

@ -0,0 +1,22 @@
#ifndef GEN4_SOURCE_H
#define GEN4_SOURCE_H
#include "compiler.h"
#include "sna.h"
#include "sna_render.h"
bool
gen4_channel_init_solid(struct sna *sna,
struct sna_composite_channel *channel,
uint32_t color);
bool
gen4_channel_init_linear(struct sna *sna,
PicturePtr picture,
struct sna_composite_channel *channel,
int x, int y,
int w, int h,
int dst_x, int dst_y);
#endif /* GEN4_SOURCE_H */

304
drivers/video/Intel-2D/gen4_vertex.c Normal file
View File

@ -0,0 +1,304 @@
/*
* Copyright © 2012 Intel Corporation
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the "Software"),
* to deal in the Software without restriction, including without limitation
* the rights to use, copy, modify, merge, publish, distribute, sublicense,
* and/or sell copies of the Software, and to permit persons to whom the
* Software is furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice (including the next
* paragraph) shall be included in all copies or substantial portions of the
* Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
* THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
* SOFTWARE.
*
* Authors:
* Chris Wilson <chris@chris-wilson.co.uk>
*
*/
#ifdef HAVE_CONFIG_H
#include "config.h"
#endif
#include "sna.h"
#include "sna_render.h"
#include "sna_render_inline.h"
#include "gen4_vertex.h"
void gen4_vertex_flush(struct sna *sna)
{
DBG(("%s[%x] = %d\n", __FUNCTION__,
4*sna->render.vertex_offset,
sna->render.vertex_index - sna->render.vertex_start));
assert(sna->render.vertex_offset);
assert(sna->render.vertex_index > sna->render.vertex_start);
sna->kgem.batch[sna->render.vertex_offset] =
sna->render.vertex_index - sna->render.vertex_start;
sna->render.vertex_offset = 0;
}
int gen4_vertex_finish(struct sna *sna)
{
struct kgem_bo *bo;
unsigned int i;
unsigned hint, size;
DBG(("%s: used=%d / %d\n", __FUNCTION__,
sna->render.vertex_used, sna->render.vertex_size));
assert(sna->render.vertex_offset == 0);
assert(sna->render.vertex_used);
sna_vertex_wait__locked(&sna->render);
/* Note: we only need dword alignment (currently) */
bo = sna->render.vbo;
if (bo) {
for (i = 0; i < sna->render.nvertex_reloc; i++) {
DBG(("%s: reloc[%d] = %d\n", __FUNCTION__,
i, sna->render.vertex_reloc[i]));
sna->kgem.batch[sna->render.vertex_reloc[i]] =
kgem_add_reloc(&sna->kgem,
sna->render.vertex_reloc[i], bo,
I915_GEM_DOMAIN_VERTEX << 16,
0);
}
assert(!sna->render.active);
sna->render.nvertex_reloc = 0;
sna->render.vertex_used = 0;
sna->render.vertex_index = 0;
sna->render.vbo = NULL;
sna->render.vb_id = 0;
kgem_bo_destroy(&sna->kgem, bo);
}
hint = CREATE_GTT_MAP;
if (bo)
hint |= CREATE_CACHED | CREATE_NO_THROTTLE;
size = 256*1024;
assert(!sna->render.active);
sna->render.vertices = NULL;
sna->render.vbo = kgem_create_linear(&sna->kgem, size, hint);
while (sna->render.vbo == NULL && size > 16*1024) {
size /= 2;
sna->render.vbo = kgem_create_linear(&sna->kgem, size, hint);
}
if (sna->render.vbo == NULL)
sna->render.vbo = kgem_create_linear(&sna->kgem,
256*1024, CREATE_GTT_MAP);
if (sna->render.vbo)
sna->render.vertices = kgem_bo_map(&sna->kgem, sna->render.vbo);
if (sna->render.vertices == NULL) {
if (sna->render.vbo) {
kgem_bo_destroy(&sna->kgem, sna->render.vbo);
sna->render.vbo = NULL;
}
sna->render.vertices = sna->render.vertex_data;
sna->render.vertex_size = ARRAY_SIZE(sna->render.vertex_data);
return 0;
}
if (sna->render.vertex_used) {
DBG(("%s: copying initial buffer x %d to handle=%d\n",
__FUNCTION__,
sna->render.vertex_used,
sna->render.vbo->handle));
assert(sizeof(float)*sna->render.vertex_used <=
__kgem_bo_size(sna->render.vbo));
memcpy(sna->render.vertices,
sna->render.vertex_data,
sizeof(float)*sna->render.vertex_used);
}
size = __kgem_bo_size(sna->render.vbo)/4;
if (size >= UINT16_MAX)
size = UINT16_MAX - 1;
DBG(("%s: create vbo handle=%d, size=%d\n",
__FUNCTION__, sna->render.vbo->handle, size));
sna->render.vertex_size = size;
return sna->render.vertex_size - sna->render.vertex_used;
}
void gen4_vertex_close(struct sna *sna)
{
struct kgem_bo *bo, *free_bo = NULL;
unsigned int i, delta = 0;
assert(sna->render.vertex_offset == 0);
if (!sna->render.vb_id)
return;
DBG(("%s: used=%d, vbo active? %d, vb=%x, nreloc=%d\n",
__FUNCTION__, sna->render.vertex_used, sna->render.vbo ? sna->render.vbo->handle : 0,
sna->render.vb_id, sna->render.nvertex_reloc));
assert(!sna->render.active);
bo = sna->render.vbo;
if (bo) {
if (sna->render.vertex_size - sna->render.vertex_used < 64) {
DBG(("%s: discarding vbo (full), handle=%d\n", __FUNCTION__, sna->render.vbo->handle));
sna->render.vbo = NULL;
sna->render.vertices = sna->render.vertex_data;
sna->render.vertex_size = ARRAY_SIZE(sna->render.vertex_data);
free_bo = bo;
} else if (IS_CPU_MAP(bo->map) && !sna->kgem.has_llc) {
DBG(("%s: converting CPU map to GTT\n", __FUNCTION__));
sna->render.vertices =
kgem_bo_map__gtt(&sna->kgem, sna->render.vbo);
if (sna->render.vertices == NULL) {
sna->render.vbo = NULL;
sna->render.vertices = sna->render.vertex_data;
sna->render.vertex_size = ARRAY_SIZE(sna->render.vertex_data);
free_bo = bo;
}
}
} else {
if (sna->kgem.nbatch + sna->render.vertex_used <= sna->kgem.surface) {
DBG(("%s: copy to batch: %d @ %d\n", __FUNCTION__,
sna->render.vertex_used, sna->kgem.nbatch));
memcpy(sna->kgem.batch + sna->kgem.nbatch,
sna->render.vertex_data,
sna->render.vertex_used * 4);
delta = sna->kgem.nbatch * 4;
bo = NULL;
sna->kgem.nbatch += sna->render.vertex_used;
} else {
bo = kgem_create_linear(&sna->kgem,
4*sna->render.vertex_used,
CREATE_NO_THROTTLE);
if (bo && !kgem_bo_write(&sna->kgem, bo,
sna->render.vertex_data,
4*sna->render.vertex_used)) {
kgem_bo_destroy(&sna->kgem, bo);
bo = NULL;
}
DBG(("%s: new vbo: %d\n", __FUNCTION__,
sna->render.vertex_used));
free_bo = bo;
}
}
assert(sna->render.nvertex_reloc);
for (i = 0; i < sna->render.nvertex_reloc; i++) {
DBG(("%s: reloc[%d] = %d\n", __FUNCTION__,
i, sna->render.vertex_reloc[i]));
sna->kgem.batch[sna->render.vertex_reloc[i]] =
kgem_add_reloc(&sna->kgem,
sna->render.vertex_reloc[i], bo,
I915_GEM_DOMAIN_VERTEX << 16,
delta);
}
sna->render.nvertex_reloc = 0;
sna->render.vb_id = 0;
if (sna->render.vbo == NULL) {
assert(!sna->render.active);
sna->render.vertex_used = 0;
sna->render.vertex_index = 0;
assert(sna->render.vertices == sna->render.vertex_data);
assert(sna->render.vertex_size == ARRAY_SIZE(sna->render.vertex_data));
}
if (free_bo)
kgem_bo_destroy(&sna->kgem, free_bo);
}
fastcall static void
emit_primitive_identity_source_mask(struct sna *sna,
const struct sna_composite_op *op,
const struct sna_composite_rectangles *r)
{
union {
struct sna_coordinate p;
float f;
} dst;
float src_x, src_y;
float msk_x, msk_y;
float w, h;
float *v;
src_x = r->src.x + op->src.offset[0];
src_y = r->src.y + op->src.offset[1];
msk_x = r->mask.x + op->mask.offset[0];
msk_y = r->mask.y + op->mask.offset[1];
w = r->width;
h = r->height;
assert(op->floats_per_rect == 15);
assert((sna->render.vertex_used % 5) == 0);
v = sna->render.vertices + sna->render.vertex_used;
sna->render.vertex_used += 15;
dst.p.x = r->dst.x + r->width;
dst.p.y = r->dst.y + r->height;
v[0] = dst.f;
v[1] = (src_x + w) * op->src.scale[0];
v[2] = (src_y + h) * op->src.scale[1];
v[3] = (msk_x + w) * op->mask.scale[0];
v[4] = (msk_y + h) * op->mask.scale[1];
dst.p.x = r->dst.x;
v[5] = dst.f;
v[6] = src_x * op->src.scale[0];
v[7] = v[2];
v[8] = msk_x * op->mask.scale[0];
v[9] = v[4];
dst.p.y = r->dst.y;
v[10] = dst.f;
v[11] = v[6];
v[12] = src_y * op->src.scale[1];
v[13] = v[8];
v[14] = msk_y * op->mask.scale[1];
}
unsigned gen4_choose_composite_emitter(struct sna_composite_op *tmp)
{
unsigned vb;
if (tmp->mask.bo) {
if (tmp->mask.transform == NULL) {
if (tmp->src.is_solid) {
DBG(("%s: solid, identity mask\n", __FUNCTION__));
} else if (tmp->src.is_linear) {
DBG(("%s: linear, identity mask\n", __FUNCTION__));
} else if (tmp->src.transform == NULL) {
DBG(("%s: identity source, identity mask\n", __FUNCTION__));
tmp->prim_emit = emit_primitive_identity_source_mask;
tmp->floats_per_vertex = 5;
vb = 2 << 2 | 2;
} else if (tmp->src.is_affine) {
DBG(("%s: simple src, identity mask\n", __FUNCTION__));
} else {
DBG(("%s: projective source, identity mask\n", __FUNCTION__));
}
} else {
DBG(("%s: general mask: floats-per-vertex=%d, vb=%x\n",
__FUNCTION__,tmp->floats_per_vertex, vb));
}
} else {
}
tmp->floats_per_rect = 3 * tmp->floats_per_vertex;
return vb;
}

1411
drivers/video/Intel-2D/gen5_render.c Normal file
View File

File diff suppressed because it is too large Load Diff

2772
drivers/video/Intel-2D/gen5_render.h Normal file
View File

File diff suppressed because it is too large Load Diff

10
drivers/video/Intel-2D/gen6_render.c
View File

@ -2683,8 +2683,8 @@ fallback:
}
bool
gen6_composite(struct sna *sna,
static bool
gen6_blit_tex(struct sna *sna,
uint8_t op,
PixmapPtr src, struct kgem_bo *src_bo,
PixmapPtr mask,struct kgem_bo *mask_bo,
@ -2722,7 +2722,6 @@ gen6_composite(struct sna *sna,
tmp->has_component_alpha = false;
tmp->need_magic_ca_pass = false;
tmp->mask.repeat = SAMPLER_EXTEND_NONE;
tmp->mask.filter = SAMPLER_FILTER_NEAREST;
tmp->mask.is_affine = true;
@ -3479,7 +3478,10 @@ bool gen6_render_init(struct sna *sna)
// sna->render.video = gen6_render_video;
// sna->render.copy_boxes = gen6_render_copy_boxes;
sna->render.copy = gen6_render_copy;
sna->render.blit_tex = gen6_blit_tex;
// sna->render.copy = gen6_render_copy;
// sna->render.fill_boxes = gen6_render_fill_boxes;
// sna->render.fill = gen6_render_fill;

1704
drivers/video/Intel-2D/gen7_render.c Normal file
View File

File diff suppressed because it is too large Load Diff

1364
drivers/video/Intel-2D/gen7_render.h Normal file
View File

File diff suppressed because it is too large Load Diff

13
drivers/video/Intel-2D/sna.c
View File

@ -113,16 +113,16 @@ bool sna_accel_init(struct sna *sna)
no_render_init(sna);
if (sna->info->gen >= 0100) {
/* } else if (sna->info->gen >= 070) {
} else if (sna->info->gen >= 070) {
if (gen7_render_init(sna))
backend = "IvyBridge"; */
backend = "IvyBridge";
} else if (sna->info->gen >= 060) {
if (gen6_render_init(sna))
backend = "SandyBridge";
/* } else if (sna->info->gen >= 050) {
} else if (sna->info->gen >= 050) {
if (gen5_render_init(sna))
backend = "Ironlake";
} else if (sna->info->gen >= 040) {
/* } else if (sna->info->gen >= 040) {
if (gen4_render_init(sna))
backend = "Broadwater/Crestline";
} else if (sna->info->gen >= 030) {
@ -334,7 +334,6 @@ done:
return kgem_bo_reference(cache->bo[i]);
}
#endif
int sna_blit_copy(bitmap_t *src_bitmap, int dst_x, int dst_y,
@ -381,6 +380,8 @@ int sna_blit_copy(bitmap_t *src_bitmap, int dst_x, int dst_y,
// __asm__ __volatile__("int3");
};
#endif
int sna_create_bitmap(bitmap_t *bitmap)
{
@ -524,7 +525,7 @@ int sna_blit_tex(bitmap_t *src_bitmap, int dst_x, int dst_y,
src_bo = (struct kgem_bo*)src_bitmap->handle;
if( gen6_composite(sna_device, PictOpSrc,
if( sna_device->render.blit_tex(sna_device, PictOpSrc,
&src, src_bo,
&mask, mask_bo,
&dst, sna_fb.fb_bo,

11
drivers/video/Intel-2D/sna.h
View File

@ -320,9 +320,9 @@ struct sna {
} blt_state;
union {
// struct gen2_render_state gen2;
// struct gen3_render_state gen3;
// struct gen4_render_state gen4;
// struct gen5_render_state gen5;
struct gen3_render_state gen3;
struct gen4_render_state gen4;
struct gen5_render_state gen5;
struct gen6_render_state gen6;
struct gen7_render_state gen7;
} render_state;
@ -347,6 +347,11 @@ struct sna {
#endif
};
static inline struct sna *
to_sna_from_kgem(struct kgem *kgem)
{
return container_of(kgem, struct sna, kgem);
}
#ifndef ARRAY_SIZE
#define ARRAY_SIZE(x) (sizeof(x) / sizeof(x[0]))

11
drivers/video/Intel-2D/sna_render.h
View File

@ -217,6 +217,17 @@ struct sna_render {
#endif
bool (*blit_tex)(struct sna *sna,
uint8_t op,
PixmapPtr src, struct kgem_bo *src_bo,
PixmapPtr mask,struct kgem_bo *mask_bo,
PixmapPtr dst, struct kgem_bo *dst_bo,
int32_t src_x, int32_t src_y,
int32_t msk_x, int32_t msk_y,
int32_t dst_x, int32_t dst_y,
int32_t width, int32_t height,
struct sna_composite_op *tmp);
bool (*copy)(struct sna *sna, uint8_t alu,
PixmapPtr src, struct kgem_bo *src_bo,
PixmapPtr dst, struct kgem_bo *dst_bo,

24
drivers/video/Intel-2D/sna_stream.c
View File

@ -93,6 +93,30 @@ struct kgem_bo *sna_static_stream_fini(struct sna *sna,
return bo;
}
unsigned
sna_static_stream_compile_sf(struct sna *sna,
struct sna_static_stream *stream,
bool (*compile)(struct brw_compile *))
{
struct brw_compile p;
brw_compile_init(&p, sna->kgem.gen,
sna_static_stream_map(stream,
64*sizeof(uint32_t), 64));
if (!compile(&p)) {
stream->used -= 64*sizeof(uint32_t);
return 0;
}
assert(p.nr_insn*sizeof(struct brw_instruction) <= 64*sizeof(uint32_t));
stream->used -= 64*sizeof(uint32_t) - p.nr_insn*sizeof(struct brw_instruction);
return sna_static_stream_offsetof(stream, p.store);
}
unsigned
sna_static_stream_compile_wm(struct sna *sna,
struct sna_static_stream *stream,