radeon_print(RADEON_STATE, RADEON_VERBOSE, "%s\n", __func__);
for (i = 0; i < R700_TEXTURE_NUMBERUNITS; i++) {
- radeonTexObj *t = r700->textures[i];
- if (t) {
- if (!t->image_override)
- bo = t->mt->bo;
- else
- bo = t->bo;
- if (bo) {
-
- r700SyncSurf(context, bo,
- RADEON_GEM_DOMAIN_GTT|RADEON_GEM_DOMAIN_VRAM,
- 0, TC_ACTION_ENA_bit);
-
- BEGIN_BATCH_NO_AUTOSTATE(9 + 4);
- R600_OUT_BATCH(CP_PACKET3(R600_IT_SET_RESOURCE, 7));
- 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_RESOURCE3);
- R600_OUT_BATCH(r700->textures[i]->SQ_TEX_RESOURCE4);
- R600_OUT_BATCH(r700->textures[i]->SQ_TEX_RESOURCE5);
- R600_OUT_BATCH(r700->textures[i]->SQ_TEX_RESOURCE6);
- R600_OUT_BATCH_RELOC(r700->textures[i]->SQ_TEX_RESOURCE2,
- bo,
- 0,
- RADEON_GEM_DOMAIN_GTT|RADEON_GEM_DOMAIN_VRAM, 0, 0);
- R600_OUT_BATCH_RELOC(r700->textures[i]->SQ_TEX_RESOURCE3,
- bo,
- r700->textures[i]->SQ_TEX_RESOURCE3,
- RADEON_GEM_DOMAIN_GTT|RADEON_GEM_DOMAIN_VRAM, 0, 0);
- END_BATCH();
- COMMIT_BATCH();
+ if (ctx->Texture.Unit[i]._ReallyEnabled) {
+ radeonTexObj *t = r700->textures[i];
+ uint32_t offset;
+ if (t) {
+ if (!t->image_override) {
+ bo = t->mt->bo;
+ offset = get_base_teximage_offset(t);
+ } else {
+ bo = t->bo;
+ offset = 0;
+ }
+ if (bo) {
+
+ r700SyncSurf(context, bo,
+ RADEON_GEM_DOMAIN_GTT|RADEON_GEM_DOMAIN_VRAM,
+ 0, TC_ACTION_ENA_bit);
+
+ BEGIN_BATCH_NO_AUTOSTATE(9 + 4);
+ R600_OUT_BATCH(CP_PACKET3(R600_IT_SET_RESOURCE, 7));
+ 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_RESOURCE3);
+ R600_OUT_BATCH(r700->textures[i]->SQ_TEX_RESOURCE4);
+ R600_OUT_BATCH(r700->textures[i]->SQ_TEX_RESOURCE5);
+ R600_OUT_BATCH(r700->textures[i]->SQ_TEX_RESOURCE6);
+ R600_OUT_BATCH_RELOC(r700->textures[i]->SQ_TEX_RESOURCE2,
+ bo,
+ offset,
+ RADEON_GEM_DOMAIN_GTT|RADEON_GEM_DOMAIN_VRAM, 0, 0);
+ R600_OUT_BATCH_RELOC(r700->textures[i]->SQ_TEX_RESOURCE3,
+ bo,
+ r700->textures[i]->SQ_TEX_RESOURCE3,
+ RADEON_GEM_DOMAIN_GTT|RADEON_GEM_DOMAIN_VRAM, 0, 0);
+ END_BATCH();
+ COMMIT_BATCH();
+ }
}
}
}
radeon_print(RADEON_STATE, RADEON_VERBOSE, "%s\n", __func__);
for (i = 0; i < R700_TEXTURE_NUMBERUNITS; i++) {
- 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);
- 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);
- END_BATCH();
- COMMIT_BATCH();
+ 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);
+ 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);
+ END_BATCH();
+ COMMIT_BATCH();
+ }
}
}
}
radeon_print(RADEON_STATE, RADEON_VERBOSE, "%s\n", __func__);
for (i = 0; i < R700_TEXTURE_NUMBERUNITS; i++) {
- radeonTexObj *t = r700->textures[i];
- if (t) {
- BEGIN_BATCH_NO_AUTOSTATE(2 + 4);
- R600_OUT_BATCH_REGSEQ((TD_PS_SAMPLER0_BORDER_RED + (i * 16)), 4);
- R600_OUT_BATCH(r700->textures[i]->TD_PS_SAMPLER0_BORDER_RED);
- R600_OUT_BATCH(r700->textures[i]->TD_PS_SAMPLER0_BORDER_GREEN);
- R600_OUT_BATCH(r700->textures[i]->TD_PS_SAMPLER0_BORDER_BLUE);
- R600_OUT_BATCH(r700->textures[i]->TD_PS_SAMPLER0_BORDER_ALPHA);
- END_BATCH();
- COMMIT_BATCH();
+ if (ctx->Texture.Unit[i]._ReallyEnabled) {
+ radeonTexObj *t = r700->textures[i];
+ if (t) {
+ BEGIN_BATCH_NO_AUTOSTATE(2 + 4);
+ R600_OUT_BATCH_REGSEQ((TD_PS_SAMPLER0_BORDER_RED + (i * 16)), 4);
+ R600_OUT_BATCH(r700->textures[i]->TD_PS_SAMPLER0_BORDER_RED);
+ R600_OUT_BATCH(r700->textures[i]->TD_PS_SAMPLER0_BORDER_GREEN);
+ R600_OUT_BATCH(r700->textures[i]->TD_PS_SAMPLER0_BORDER_BLUE);
+ R600_OUT_BATCH(r700->textures[i]->TD_PS_SAMPLER0_BORDER_ALPHA);
+ END_BATCH();
+ COMMIT_BATCH();
+ }
}
}
}
+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 */
+ StreamDesc * pStreamDesc)
{
context_t *context = R700_CONTEXT(ctx);
struct radeon_aos * paos = (struct radeon_aos *)pAos;
+ unsigned int nVBsize;
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;
else
r700SyncSurf(context, paos->bo, RADEON_GEM_DOMAIN_GTT, 0, VC_ACTION_ENA_bit);
+ if(0 == pStreamDesc->stride)
+ {
+ nVBsize = paos->count * pStreamDesc->size * getTypeSize(pStreamDesc->type);
+ }
+ else
+ {
+ nVBsize = paos->count * pStreamDesc->stride;
+ }
+
uSQ_VTX_CONSTANT_WORD0_0 = paos->offset;
- uSQ_VTX_CONSTANT_WORD1_0 = count * (size * 4) - 1;
+ uSQ_VTX_CONSTANT_WORD1_0 = nVBsize - 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,
+ SETfield(uSQ_VTX_CONSTANT_WORD2_0, pStreamDesc->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),
+ SETfield(uSQ_VTX_CONSTANT_WORD2_0, GetSurfaceFormat(pStreamDesc->type, pStreamDesc->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);
+
+ if(GL_TRUE == pStreamDesc->normalize)
+ {
+ 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);
+ //}
+
+ if(1 == pStreamDesc->_signed)
+ {
+ 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,
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((pStreamDesc->element + 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);
}
-void r700SetupStreams(GLcontext *ctx)
-{
- context_t *context = R700_CONTEXT(ctx);
- struct r700_vertex_program *vpc
- = (struct r700_vertex_program *)ctx->VertexProgram._Current;
- 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(vpc->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);
- struct r700_vertex_program *vpc
- = (struct r700_vertex_program *)ctx->VertexProgram._Current;
+ struct r700_vertex_program *vp = context->selected_vp;
unsigned int i, j = 0;
BATCH_LOCALS(&context->radeon);
radeon_print(RADEON_STATE, RADEON_VERBOSE, "%s\n", __func__);
COMMIT_BATCH();
for(i=0; i<VERT_ATTRIB_MAX; i++) {
- if(vpc->mesa_program.Base.InputsRead & (1 << i)) {
- /* 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);
- j++;
+ if(vp->mesa_program->Base.InputsRead & (1 << i))
+ {
+ r700SetupVTXConstants(ctx,
+ (void*)(&context->radeon.tcl.aos[j]),
+ &(context->stream_desc[j]));
+ j++;
}
}
}
SETfield(r700->DB_DEPTH_INFO.u32All, DEPTH_16,
DB_DEPTH_INFO__FORMAT_shift, DB_DEPTH_INFO__FORMAT_mask);
}
- SETfield(r700->DB_DEPTH_INFO.u32All, ARRAY_2D_TILED_THIN1,
+ SETfield(r700->DB_DEPTH_INFO.u32All, ARRAY_1D_TILED_THIN1,
DB_DEPTH_INFO__ARRAY_MODE_shift, DB_DEPTH_INFO__ARRAY_MODE_mask);
/* r700->DB_PREFETCH_LIMIT.bits.DEPTH_HEIGHT_TILE_MAX = (context->currentDraw->h >> 3) - 1; */ /* z buffer sie may much bigger than what need, so use actual used h. */
}
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;
}
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;
R700_CHIP_CONTEXT *r700 = (R700_CHIP_CONTEXT*)(&context->hw);
for (i = 0; i < R700_TEXTURE_NUMBERUNITS; i++) {
- radeonTexObj *t = r700->textures[i];
- if (t)
- count++;
+ if (ctx->Texture.Unit[i]._ReallyEnabled) {
+ radeonTexObj *t = r700->textures[i];
+ if (t)
+ count++;
+ }
}
radeon_print(RADEON_STATE, RADEON_TRACE, "%s %d\n", __func__, count);
return count * 31;
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(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;