#include "r600_context.h"
#include "r600_cmdbuf.h"
-#include "r700_state.h"
#include "r600_tex.h"
#include "r700_oglprog.h"
#include "r700_fragprog.h"
#include "r700_vertprog.h"
-#include "r700_ioctl.h"
#include "radeon_mipmap_tree.h"
{
context_t *context = R700_CONTEXT(ctx);
R700_CHIP_CONTEXT *r700 = (R700_CHIP_CONTEXT*)(&context->hw);
+
+ struct r700_vertex_program *vp = context->selected_vp;
+
struct radeon_bo *bo = NULL;
unsigned int i;
BATCH_LOCALS(&context->radeon);
radeon_print(RADEON_STATE, RADEON_VERBOSE, "%s\n", __func__);
for (i = 0; i < R700_TEXTURE_NUMBERUNITS; i++) {
- if (ctx->Texture.Unit[i]._ReallyEnabled) {
+ if (ctx->Texture.Unit[i]._ReallyEnabled) {
radeonTexObj *t = r700->textures[i];
if (t) {
- if (!t->image_override)
+ if (!t->image_override) {
bo = t->mt->bo;
- else
+ } else {
bo = t->bo;
+ }
if (bo) {
r700SyncSurf(context, bo,
BEGIN_BATCH_NO_AUTOSTATE(9 + 4);
R600_OUT_BATCH(CP_PACKET3(R600_IT_SET_RESOURCE, 7));
- R600_OUT_BATCH(i * 7);
+
+ if( (1<<i) & vp->r700AsmCode.unVetTexBits )
+ { /* vs texture */
+ R600_OUT_BATCH((i + VERT_ATTRIB_MAX + SQ_FETCH_RESOURCE_VS_OFFSET) * FETCH_RESOURCE_STRIDE);
+ }
+ else
+ {
+ R600_OUT_BATCH(i * 7);
+ }
+
R600_OUT_BATCH(r700->textures[i]->SQ_TEX_RESOURCE0);
R600_OUT_BATCH(r700->textures[i]->SQ_TEX_RESOURCE1);
R600_OUT_BATCH(r700->textures[i]->SQ_TEX_RESOURCE2);
R600_OUT_BATCH(r700->textures[i]->SQ_TEX_RESOURCE6);
R600_OUT_BATCH_RELOC(r700->textures[i]->SQ_TEX_RESOURCE2,
bo,
- 0,
+ r700->textures[i]->SQ_TEX_RESOURCE2,
RADEON_GEM_DOMAIN_GTT|RADEON_GEM_DOMAIN_VRAM, 0, 0);
R600_OUT_BATCH_RELOC(r700->textures[i]->SQ_TEX_RESOURCE3,
bo,
}
}
+#define SAMPLER_STRIDE 3
+
static void r700SendTexSamplerState(GLcontext *ctx, struct radeon_state_atom *atom)
{
context_t *context = R700_CONTEXT(ctx);
R700_CHIP_CONTEXT *r700 = (R700_CHIP_CONTEXT*)(&context->hw);
unsigned int i;
+
+ struct r700_vertex_program *vp = context->selected_vp;
+
BATCH_LOCALS(&context->radeon);
radeon_print(RADEON_STATE, RADEON_VERBOSE, "%s\n", __func__);
for (i = 0; i < R700_TEXTURE_NUMBERUNITS; i++) {
- if (ctx->Texture.Unit[i]._ReallyEnabled) {
+ if (ctx->Texture.Unit[i]._ReallyEnabled) {
radeonTexObj *t = r700->textures[i];
if (t) {
BEGIN_BATCH_NO_AUTOSTATE(5);
R600_OUT_BATCH(CP_PACKET3(R600_IT_SET_SAMPLER, 3));
- R600_OUT_BATCH(i * 3);
+
+ if( (1<<i) & vp->r700AsmCode.unVetTexBits )
+ { /* vs texture */
+ R600_OUT_BATCH((i+SQ_TEX_SAMPLER_VS_OFFSET) * SAMPLER_STRIDE); //work 1
+ }
+ else
+ {
+ R600_OUT_BATCH(i * 3);
+ }
+
R600_OUT_BATCH(r700->textures[i]->SQ_TEX_SAMPLER0);
R600_OUT_BATCH(r700->textures[i]->SQ_TEX_SAMPLER1);
R600_OUT_BATCH(r700->textures[i]->SQ_TEX_SAMPLER2);
}
}
+extern int getTypeSize(GLenum type);
static void r700SetupVTXConstants(GLcontext * ctx,
- unsigned int nStreamID,
void * pAos,
- unsigned int size, /* number of elements in vector */
- unsigned int stride,
- unsigned int count) /* number of vectors in stream */
-{
- context_t *context = R700_CONTEXT(ctx);
- struct radeon_aos * paos = (struct radeon_aos *)pAos;
- BATCH_LOCALS(&context->radeon);
- radeon_print(RADEON_STATE, RADEON_VERBOSE, "%s\n", __func__);
-
- unsigned int uSQ_VTX_CONSTANT_WORD0_0;
- unsigned int uSQ_VTX_CONSTANT_WORD1_0;
- unsigned int uSQ_VTX_CONSTANT_WORD2_0 = 0;
- unsigned int uSQ_VTX_CONSTANT_WORD3_0 = 0;
- unsigned int uSQ_VTX_CONSTANT_WORD6_0 = 0;
-
- if (!paos->bo)
- return;
-
- if ((context->radeon.radeonScreen->chip_family == CHIP_FAMILY_RV610) ||
- (context->radeon.radeonScreen->chip_family == CHIP_FAMILY_RV620) ||
- (context->radeon.radeonScreen->chip_family == CHIP_FAMILY_RS780) ||
- (context->radeon.radeonScreen->chip_family == CHIP_FAMILY_RS880) ||
- (context->radeon.radeonScreen->chip_family == CHIP_FAMILY_RV710))
- r700SyncSurf(context, paos->bo, RADEON_GEM_DOMAIN_GTT, 0, TC_ACTION_ENA_bit);
- else
- r700SyncSurf(context, paos->bo, RADEON_GEM_DOMAIN_GTT, 0, VC_ACTION_ENA_bit);
-
- uSQ_VTX_CONSTANT_WORD0_0 = paos->offset;
- uSQ_VTX_CONSTANT_WORD1_0 = count * (size * 4) - 1;
-
- SETfield(uSQ_VTX_CONSTANT_WORD2_0, 0, BASE_ADDRESS_HI_shift, BASE_ADDRESS_HI_mask); /* TODO */
- SETfield(uSQ_VTX_CONSTANT_WORD2_0, stride, SQ_VTX_CONSTANT_WORD2_0__STRIDE_shift,
- SQ_VTX_CONSTANT_WORD2_0__STRIDE_mask);
- SETfield(uSQ_VTX_CONSTANT_WORD2_0, GetSurfaceFormat(GL_FLOAT, size, NULL),
- SQ_VTX_CONSTANT_WORD2_0__DATA_FORMAT_shift,
- SQ_VTX_CONSTANT_WORD2_0__DATA_FORMAT_mask); /* TODO : trace back api for initial data type, not only GL_FLOAT */
- SETfield(uSQ_VTX_CONSTANT_WORD2_0, SQ_NUM_FORMAT_SCALED,
- SQ_VTX_CONSTANT_WORD2_0__NUM_FORMAT_ALL_shift, SQ_VTX_CONSTANT_WORD2_0__NUM_FORMAT_ALL_mask);
- SETbit(uSQ_VTX_CONSTANT_WORD2_0, SQ_VTX_CONSTANT_WORD2_0__FORMAT_COMP_ALL_bit);
-
- SETfield(uSQ_VTX_CONSTANT_WORD3_0, 1, MEM_REQUEST_SIZE_shift, MEM_REQUEST_SIZE_mask);
- SETfield(uSQ_VTX_CONSTANT_WORD6_0, SQ_TEX_VTX_VALID_BUFFER,
- SQ_TEX_RESOURCE_WORD6_0__TYPE_shift, SQ_TEX_RESOURCE_WORD6_0__TYPE_mask);
-
- BEGIN_BATCH_NO_AUTOSTATE(9 + 2);
-
- R600_OUT_BATCH(CP_PACKET3(R600_IT_SET_RESOURCE, 7));
- R600_OUT_BATCH((nStreamID + SQ_FETCH_RESOURCE_VS_OFFSET) * FETCH_RESOURCE_STRIDE);
- R600_OUT_BATCH(uSQ_VTX_CONSTANT_WORD0_0);
- R600_OUT_BATCH(uSQ_VTX_CONSTANT_WORD1_0);
- R600_OUT_BATCH(uSQ_VTX_CONSTANT_WORD2_0);
- R600_OUT_BATCH(uSQ_VTX_CONSTANT_WORD3_0);
- R600_OUT_BATCH(0);
- R600_OUT_BATCH(0);
- R600_OUT_BATCH(uSQ_VTX_CONSTANT_WORD6_0);
- R600_OUT_BATCH_RELOC(uSQ_VTX_CONSTANT_WORD0_0,
- paos->bo,
- uSQ_VTX_CONSTANT_WORD0_0,
- RADEON_GEM_DOMAIN_GTT, 0, 0);
- END_BATCH();
- COMMIT_BATCH();
-
-}
-
-extern int getTypeSize(GLenum type);
-static void r700SetupVTXConstants2(GLcontext * ctx,
- void * pAos,
- StreamDesc * pStreamDesc)
+ StreamDesc * pStreamDesc)
{
context_t *context = R700_CONTEXT(ctx);
struct radeon_aos * paos = (struct radeon_aos *)pAos;
}
else
{
- nVBsize = paos->count * pStreamDesc->stride;
+ nVBsize = (paos->count - 1) * pStreamDesc->stride
+ + pStreamDesc->size * getTypeSize(pStreamDesc->type);
}
uSQ_VTX_CONSTANT_WORD0_0 = paos->offset;
SETfield(uSQ_VTX_CONSTANT_WORD2_0, SQ_NUM_FORMAT_NORM,
SQ_VTX_CONSTANT_WORD2_0__NUM_FORMAT_ALL_shift, SQ_VTX_CONSTANT_WORD2_0__NUM_FORMAT_ALL_mask);
}
- //else
- //{
- // SETfield(uSQ_VTX_CONSTANT_WORD2_0, SQ_NUM_FORMAT_INT,
- // SQ_VTX_CONSTANT_WORD2_0__NUM_FORMAT_ALL_shift, SQ_VTX_CONSTANT_WORD2_0__NUM_FORMAT_ALL_mask);
- //}
+ else
+ {
+ SETfield(uSQ_VTX_CONSTANT_WORD2_0, SQ_NUM_FORMAT_SCALED,
+ SQ_VTX_CONSTANT_WORD2_0__NUM_FORMAT_ALL_shift, SQ_VTX_CONSTANT_WORD2_0__NUM_FORMAT_ALL_mask);
+ }
if(1 == pStreamDesc->_signed)
{
}
-void r700SetupStreams(GLcontext *ctx)
-{
- context_t *context = R700_CONTEXT(ctx);
- struct r700_vertex_program *vp = context->selected_vp;
- TNLcontext *tnl = TNL_CONTEXT(ctx);
- struct vertex_buffer *vb = &tnl->vb;
- unsigned int i, j = 0;
- radeon_print(RADEON_STATE, RADEON_VERBOSE, "%s\n", __func__);
-
- R600_STATECHANGE(context, vtx);
-
- for(i=0; i<VERT_ATTRIB_MAX; i++) {
- if(vp->mesa_program->Base.InputsRead & (1 << i)) {
- rcommon_emit_vector(ctx,
- &context->radeon.tcl.aos[j],
- vb->AttribPtr[i]->data,
- vb->AttribPtr[i]->size,
- vb->AttribPtr[i]->stride,
- vb->Count);
- j++;
- }
- }
- context->radeon.tcl.aos_count = j;
-}
-
static void r700SendVTXState(GLcontext *ctx, struct radeon_state_atom *atom)
{
context_t *context = R700_CONTEXT(ctx);
COMMIT_BATCH();
for(i=0; i<VERT_ATTRIB_MAX; i++) {
- if(vp->mesa_program->Base.InputsRead & (1 << i))
- {
- if(1 == context->selected_vp->uiVersion)
- {
- /* currently aos are packed */
- r700SetupVTXConstants(ctx,
- i,
- (void*)(&context->radeon.tcl.aos[j]),
- (unsigned int)context->radeon.tcl.aos[j].components,
- (unsigned int)context->radeon.tcl.aos[j].stride * 4,
- (unsigned int)context->radeon.tcl.aos[j].count);
- }
- else
- { /* context->selected_vp->uiVersion == 2 : aos not always packed */
- r700SetupVTXConstants2(ctx,
- (void*)(&context->radeon.tcl.aos[j]),
- &(context->stream_desc[j]));
- }
- j++;
+ if(vp->mesa_program->Base.InputsRead & (1 << i))
+ {
+ r700SetupVTXConstants(ctx,
+ (void*)(&context->radeon.tcl.aos[j]),
+ &(context->stream_desc[j]));
+ j++;
}
}
}
R600_STATECHANGE(context, cb_target);
/* color buffer */
- r700->render_target[id].CB_COLOR0_BASE.u32All = context->radeon.state.color.draw_offset;
+ r700->render_target[id].CB_COLOR0_BASE.u32All = context->radeon.state.color.draw_offset / 256;
nPitchInPixel = rrb->pitch/rrb->cpp;
SETfield(r700->render_target[id].CB_COLOR0_SIZE.u32All, (nPitchInPixel/8)-1,
PITCH_TILE_MAX_shift, PITCH_TILE_MAX_mask);
SETfield(r700->render_target[id].CB_COLOR0_SIZE.u32All, ( (nPitchInPixel * context->radeon.radeonScreen->driScreen->fbHeight)/64 )-1,
SLICE_TILE_MAX_shift, SLICE_TILE_MAX_mask);
- r700->render_target[id].CB_COLOR0_BASE.u32All = 0;
SETfield(r700->render_target[id].CB_COLOR0_INFO.u32All, ENDIAN_NONE, ENDIAN_shift, ENDIAN_mask);
SETfield(r700->render_target[id].CB_COLOR0_INFO.u32All, ARRAY_LINEAR_GENERAL,
CB_COLOR0_INFO__ARRAY_MODE_shift, CB_COLOR0_INFO__ARRAY_MODE_mask);
rrb = radeon_get_depthbuffer(&context->radeon);
if (!rrb || !rrb->bo) {
- fprintf(stderr, "no rrb\n");
return;
}
rrb = radeon_get_colorbuffer(&context->radeon);
if (!rrb || !rrb->bo) {
- fprintf(stderr, "no rrb\n");
return;
}
R600_OUT_BATCH((2 << id));
END_BATCH();
}
+ /* Set CMASK & TILE buffer to the offset of color buffer as
+ * we don't use those this shouldn't cause any issue and we
+ * then have a valid cmd stream
+ */
+ BEGIN_BATCH_NO_AUTOSTATE(3 + 2);
+ R600_OUT_BATCH_REGSEQ(CB_COLOR0_TILE + (4 * id), 1);
+ R600_OUT_BATCH(r700->render_target[id].CB_COLOR0_TILE.u32All);
+ R600_OUT_BATCH_RELOC(r700->render_target[id].CB_COLOR0_BASE.u32All,
+ rrb->bo,
+ r700->render_target[id].CB_COLOR0_BASE.u32All,
+ 0, RADEON_GEM_DOMAIN_VRAM, 0);
+ END_BATCH();
+ BEGIN_BATCH_NO_AUTOSTATE(3 + 2);
+ R600_OUT_BATCH_REGSEQ(CB_COLOR0_FRAG + (4 * id), 1);
+ R600_OUT_BATCH(r700->render_target[id].CB_COLOR0_FRAG.u32All);
+ R600_OUT_BATCH_RELOC(r700->render_target[id].CB_COLOR0_BASE.u32All,
+ rrb->bo,
+ r700->render_target[id].CB_COLOR0_BASE.u32All,
+ 0, RADEON_GEM_DOMAIN_VRAM, 0);
+ END_BATCH();
- BEGIN_BATCH_NO_AUTOSTATE(18);
+ BEGIN_BATCH_NO_AUTOSTATE(12);
R600_OUT_BATCH_REGVAL(CB_COLOR0_SIZE + (4 * id), r700->render_target[id].CB_COLOR0_SIZE.u32All);
R600_OUT_BATCH_REGVAL(CB_COLOR0_VIEW + (4 * id), r700->render_target[id].CB_COLOR0_VIEW.u32All);
R600_OUT_BATCH_REGVAL(CB_COLOR0_INFO + (4 * id), r700->render_target[id].CB_COLOR0_INFO.u32All);
- R600_OUT_BATCH_REGVAL(CB_COLOR0_TILE + (4 * id), r700->render_target[id].CB_COLOR0_TILE.u32All);
- R600_OUT_BATCH_REGVAL(CB_COLOR0_FRAG + (4 * id), r700->render_target[id].CB_COLOR0_FRAG.u32All);
R600_OUT_BATCH_REGVAL(CB_COLOR0_MASK + (4 * id), r700->render_target[id].CB_COLOR0_MASK.u32All);
END_BATCH();
static void r700SendPSState(GLcontext *ctx, struct radeon_state_atom *atom)
{
- context_t *context = R700_CONTEXT(ctx);
- R700_CHIP_CONTEXT *r700 = R700_CONTEXT_STATES(context);
- struct radeon_bo * pbo;
- BATCH_LOCALS(&context->radeon);
- radeon_print(RADEON_STATE, RADEON_VERBOSE, "%s\n", __func__);
+ context_t *context = R700_CONTEXT(ctx);
+ R700_CHIP_CONTEXT *r700 = R700_CONTEXT_STATES(context);
+ struct radeon_bo * pbo;
+ BATCH_LOCALS(&context->radeon);
+ radeon_print(RADEON_STATE, RADEON_VERBOSE, "%s\n", __func__);
- pbo = (struct radeon_bo *)r700GetActiveFpShaderBo(GL_CONTEXT(context));
+ pbo = (struct radeon_bo *)r700GetActiveFpShaderBo(GL_CONTEXT(context));
- if (!pbo)
- return;
+ if (!pbo)
+ return;
- r700SyncSurf(context, pbo, RADEON_GEM_DOMAIN_GTT, 0, SH_ACTION_ENA_bit);
+ r700SyncSurf(context, pbo, RADEON_GEM_DOMAIN_GTT, 0, SH_ACTION_ENA_bit);
- BEGIN_BATCH_NO_AUTOSTATE(3 + 2);
- R600_OUT_BATCH_REGSEQ(SQ_PGM_START_PS, 1);
- R600_OUT_BATCH(r700->ps.SQ_PGM_START_PS.u32All);
- R600_OUT_BATCH_RELOC(r700->ps.SQ_PGM_START_PS.u32All,
- pbo,
- r700->ps.SQ_PGM_START_PS.u32All,
- RADEON_GEM_DOMAIN_GTT, 0, 0);
- END_BATCH();
+ BEGIN_BATCH_NO_AUTOSTATE(3 + 2);
+ R600_OUT_BATCH_REGSEQ(SQ_PGM_START_PS, 1);
+ R600_OUT_BATCH(r700->ps.SQ_PGM_START_PS.u32All);
+ R600_OUT_BATCH_RELOC(r700->ps.SQ_PGM_START_PS.u32All,
+ pbo,
+ r700->ps.SQ_PGM_START_PS.u32All,
+ RADEON_GEM_DOMAIN_GTT, 0, 0);
+ END_BATCH();
- BEGIN_BATCH_NO_AUTOSTATE(9);
- R600_OUT_BATCH_REGVAL(SQ_PGM_RESOURCES_PS, r700->ps.SQ_PGM_RESOURCES_PS.u32All);
- R600_OUT_BATCH_REGVAL(SQ_PGM_EXPORTS_PS, r700->ps.SQ_PGM_EXPORTS_PS.u32All);
- R600_OUT_BATCH_REGVAL(SQ_PGM_CF_OFFSET_PS, r700->ps.SQ_PGM_CF_OFFSET_PS.u32All);
- END_BATCH();
+ BEGIN_BATCH_NO_AUTOSTATE(9);
+ R600_OUT_BATCH_REGVAL(SQ_PGM_RESOURCES_PS, r700->ps.SQ_PGM_RESOURCES_PS.u32All);
+ R600_OUT_BATCH_REGVAL(SQ_PGM_EXPORTS_PS, r700->ps.SQ_PGM_EXPORTS_PS.u32All);
+ R600_OUT_BATCH_REGVAL(SQ_PGM_CF_OFFSET_PS, r700->ps.SQ_PGM_CF_OFFSET_PS.u32All);
+ END_BATCH();
- COMMIT_BATCH();
+ BEGIN_BATCH_NO_AUTOSTATE(3);
+ R600_OUT_BATCH_REGVAL(SQ_LOOP_CONST_0, 0x01000FFF);
+ END_BATCH();
+
+ COMMIT_BATCH();
}
static void r700SendVSState(GLcontext *ctx, struct radeon_state_atom *atom)
{
- context_t *context = R700_CONTEXT(ctx);
- R700_CHIP_CONTEXT *r700 = R700_CONTEXT_STATES(context);
- struct radeon_bo * pbo;
- BATCH_LOCALS(&context->radeon);
- radeon_print(RADEON_STATE, RADEON_VERBOSE, "%s\n", __func__);
+ context_t *context = R700_CONTEXT(ctx);
+ R700_CHIP_CONTEXT *r700 = R700_CONTEXT_STATES(context);
+ struct radeon_bo * pbo;
+ BATCH_LOCALS(&context->radeon);
+ radeon_print(RADEON_STATE, RADEON_VERBOSE, "%s\n", __func__);
- pbo = (struct radeon_bo *)r700GetActiveVpShaderBo(GL_CONTEXT(context));
+ pbo = (struct radeon_bo *)r700GetActiveVpShaderBo(GL_CONTEXT(context));
- if (!pbo)
- return;
+ if (!pbo)
+ return;
- r700SyncSurf(context, pbo, RADEON_GEM_DOMAIN_GTT, 0, SH_ACTION_ENA_bit);
+ r700SyncSurf(context, pbo, RADEON_GEM_DOMAIN_GTT, 0, SH_ACTION_ENA_bit);
- BEGIN_BATCH_NO_AUTOSTATE(3 + 2);
- R600_OUT_BATCH_REGSEQ(SQ_PGM_START_VS, 1);
- R600_OUT_BATCH(r700->vs.SQ_PGM_START_VS.u32All);
- R600_OUT_BATCH_RELOC(r700->vs.SQ_PGM_START_VS.u32All,
- pbo,
- r700->vs.SQ_PGM_START_VS.u32All,
- RADEON_GEM_DOMAIN_GTT, 0, 0);
- END_BATCH();
+ BEGIN_BATCH_NO_AUTOSTATE(3 + 2);
+ R600_OUT_BATCH_REGSEQ(SQ_PGM_START_VS, 1);
+ R600_OUT_BATCH(r700->vs.SQ_PGM_START_VS.u32All);
+ R600_OUT_BATCH_RELOC(r700->vs.SQ_PGM_START_VS.u32All,
+ pbo,
+ r700->vs.SQ_PGM_START_VS.u32All,
+ RADEON_GEM_DOMAIN_GTT, 0, 0);
+ END_BATCH();
- BEGIN_BATCH_NO_AUTOSTATE(6);
- R600_OUT_BATCH_REGVAL(SQ_PGM_RESOURCES_VS, r700->vs.SQ_PGM_RESOURCES_VS.u32All);
- R600_OUT_BATCH_REGVAL(SQ_PGM_CF_OFFSET_VS, r700->vs.SQ_PGM_CF_OFFSET_VS.u32All);
- END_BATCH();
+ BEGIN_BATCH_NO_AUTOSTATE(6);
+ R600_OUT_BATCH_REGVAL(SQ_PGM_RESOURCES_VS, r700->vs.SQ_PGM_RESOURCES_VS.u32All);
+ R600_OUT_BATCH_REGVAL(SQ_PGM_CF_OFFSET_VS, r700->vs.SQ_PGM_CF_OFFSET_VS.u32All);
+ END_BATCH();
- COMMIT_BATCH();
+ BEGIN_BATCH_NO_AUTOSTATE(3);
+ R600_OUT_BATCH_REGVAL((SQ_LOOP_CONST_0 + 32*4), 0x0100000F);
+ //R600_OUT_BATCH_REGVAL((SQ_LOOP_CONST_0 + (SQ_LOOP_CONST_vs<2)), 0x0100000F);
+ END_BATCH();
+
+ COMMIT_BATCH();
}
static void r700SendFSState(GLcontext *ctx, struct radeon_state_atom *atom)
BATCH_LOCALS(&context->radeon);
radeon_print(RADEON_STATE, RADEON_VERBOSE, "%s\n", __func__);
- BEGIN_BATCH_NO_AUTOSTATE(23);
- R600_OUT_BATCH_REGVAL(DB_HTILE_DATA_BASE, r700->DB_HTILE_DATA_BASE.u32All);
+ BEGIN_BATCH_NO_AUTOSTATE(17);
R600_OUT_BATCH_REGSEQ(DB_STENCIL_CLEAR, 2);
R600_OUT_BATCH(r700->DB_STENCIL_CLEAR.u32All);
R600_OUT_BATCH(r700->DB_RENDER_CONTROL.u32All);
R600_OUT_BATCH(r700->DB_RENDER_OVERRIDE.u32All);
- R600_OUT_BATCH_REGVAL(DB_HTILE_SURFACE, r700->DB_HTILE_SURFACE.u32All);
R600_OUT_BATCH_REGVAL(DB_ALPHA_TO_MASK, r700->DB_ALPHA_TO_MASK.u32All);
END_BATCH();
COMMIT_BATCH();
}
+static void r700SendQueryBegin(GLcontext *ctx, struct radeon_state_atom *atom)
+{
+ radeonContextPtr radeon = RADEON_CONTEXT(ctx);
+ struct radeon_query_object *query = radeon->query.current;
+ BATCH_LOCALS(radeon);
+ radeon_print(RADEON_STATE, RADEON_VERBOSE, "%s\n", __func__);
+
+ /* clear the buffer */
+ radeon_bo_map(query->bo, GL_FALSE);
+ memset(query->bo->ptr, 0, 4 * 2 * sizeof(uint64_t)); /* 4 DBs, 2 qwords each */
+ radeon_bo_unmap(query->bo);
+
+ radeon_cs_space_check_with_bo(radeon->cmdbuf.cs,
+ query->bo,
+ 0, RADEON_GEM_DOMAIN_GTT);
+
+ BEGIN_BATCH_NO_AUTOSTATE(4 + 2);
+ R600_OUT_BATCH(CP_PACKET3(R600_IT_EVENT_WRITE, 2));
+ R600_OUT_BATCH(ZPASS_DONE);
+ R600_OUT_BATCH(query->curr_offset); /* hw writes qwords */
+ R600_OUT_BATCH(0x00000000);
+ R600_OUT_BATCH_RELOC(VGT_EVENT_INITIATOR, query->bo, 0, 0, RADEON_GEM_DOMAIN_GTT, 0);
+ END_BATCH();
+ query->emitted_begin = GL_TRUE;
+}
+
static int check_always(GLcontext *ctx, struct radeon_state_atom *atom)
{
return atom->cmd_size;
count += 3;
if (context->radeon.radeonScreen->chip_family > CHIP_FAMILY_R600) {
- for (ui = 0; ui < R700_MAX_RENDER_TARGETS; ui++) {
+ /* targets are enabled in r700SetRenderTarget but state
+ size is calculated before that. Until MRT's are done
+ hardcode target0 as enabled. */
+ count += 3;
+ for (ui = 1; ui < R700_MAX_RENDER_TARGETS; ui++) {
if (r700->render_target[ui].enabled)
count += 3;
}
return count;
}
+static int check_queryobj(GLcontext *ctx, struct radeon_state_atom *atom)
+{
+ radeonContextPtr radeon = RADEON_CONTEXT(ctx);
+ struct radeon_query_object *query = radeon->query.current;
+ int count;
+
+ if (!query || query->emitted_begin)
+ count = 0;
+ else
+ count = atom->cmd_size;
+ radeon_print(RADEON_STATE, RADEON_TRACE, "%s %d\n", __func__, count);
+ return count;
+}
+
#define ALLOC_STATE( ATOM, CHK, SZ, EMIT ) \
do { \
context->atoms.ATOM.cmd_size = (SZ); \
insert_at_tail(&context->radeon.hw.atomlist, &context->atoms.ATOM); \
} while (0)
+static void r600_init_query_stateobj(radeonContextPtr radeon, int SZ)
+{
+ radeon->query.queryobj.cmd_size = (SZ);
+ radeon->query.queryobj.cmd = NULL;
+ radeon->query.queryobj.name = "queryobj";
+ radeon->query.queryobj.idx = 0;
+ radeon->query.queryobj.check = check_queryobj;
+ radeon->query.queryobj.dirty = GL_FALSE;
+ radeon->query.queryobj.emit = r700SendQueryBegin;
+ radeon->hw.max_state_size += (SZ);
+ insert_at_tail(&radeon->hw.atomlist, &radeon->query.queryobj);
+}
+
void r600InitAtoms(context_t *context)
{
radeon_print(RADEON_STATE, RADEON_NORMAL, "%s %p\n", __func__, context);
context->radeon.hw.atomlist.name = "atom-list";
ALLOC_STATE(sq, always, 34, r700SendSQConfig);
- ALLOC_STATE(db, always, 23, r700SendDBState);
+ ALLOC_STATE(db, always, 17, r700SendDBState);
ALLOC_STATE(stencil, always, 4, r700SendStencilState);
ALLOC_STATE(db_target, always, 12, r700SendDepthTargetState);
ALLOC_STATE(sc, always, 15, r700SendSCState);
ALLOC_STATE(poly, always, 10, r700SendPolyState);
ALLOC_STATE(cb, cb, 18, r700SendCBState);
ALLOC_STATE(clrcmp, always, 6, r700SendCBCLRCMPState);
+ ALLOC_STATE(cb_target, always, 25, r700SendRenderTargetState);
ALLOC_STATE(blnd, blnd, (6 + (R700_MAX_RENDER_TARGETS * 3)), r700SendCBBlendState);
ALLOC_STATE(blnd_clr, always, 6, r700SendCBBlendColorState);
- ALLOC_STATE(cb_target, always, 25, r700SendRenderTargetState);
ALLOC_STATE(sx, always, 9, r700SendSXState);
ALLOC_STATE(vgt, always, 41, r700SendVGTState);
ALLOC_STATE(spi, always, (59 + R700_MAX_SHADER_EXPORTS), r700SendSPIState);
ALLOC_STATE(vpt, always, 16, r700SendViewportState);
ALLOC_STATE(fs, always, 18, r700SendFSState);
- ALLOC_STATE(vs, always, 18, r700SendVSState);
- ALLOC_STATE(ps, always, 21, r700SendPSState);
+ ALLOC_STATE(vs, always, 21, r700SendVSState);
+ ALLOC_STATE(ps, always, 24, r700SendPSState);
ALLOC_STATE(vs_consts, vs_consts, (2 + (R700_MAX_DX9_CONSTS * 4)), r700SendVSConsts);
ALLOC_STATE(ps_consts, ps_consts, (2 + (R700_MAX_DX9_CONSTS * 4)), r700SendPSConsts);
ALLOC_STATE(vtx, vtx, (6 + (VERT_ATTRIB_MAX * 18)), r700SendVTXState);
ALLOC_STATE(tx, tx, (R700_TEXTURE_NUMBERUNITS * 20), r700SendTexState);
ALLOC_STATE(tx_smplr, tx, (R700_TEXTURE_NUMBERUNITS * 5), r700SendTexSamplerState);
ALLOC_STATE(tx_brdr_clr, tx, (R700_TEXTURE_NUMBERUNITS * 6), r700SendTexBorderColorState);
+ r600_init_query_stateobj(&context->radeon, 6 * 2);
context->radeon.hw.is_dirty = GL_TRUE;
context->radeon.hw.all_dirty = GL_TRUE;
projects / mesa.git / blobdiff