* IN THE SOFTWARE.
*/
#include "main/mtypes.h"
+#include "main/blend.h"
#include "main/samplerobj.h"
+#include "main/texformat.h"
#include "program/prog_parameter.h"
#include "intel_mipmap_tree.h"
return HSW_SCS_ZERO;
}
-void
-gen7_set_surface_tiling(struct gen7_surface_state *surf, uint32_t tiling)
+uint32_t
+gen7_surface_tiling_mode(uint32_t tiling)
{
switch (tiling) {
- case I915_TILING_NONE:
- surf->ss0.tiled_surface = 0;
- surf->ss0.tile_walk = 0;
- break;
case I915_TILING_X:
- surf->ss0.tiled_surface = 1;
- surf->ss0.tile_walk = BRW_TILEWALK_XMAJOR;
- break;
+ return GEN7_SURFACE_TILING_X;
case I915_TILING_Y:
- surf->ss0.tiled_surface = 1;
- surf->ss0.tile_walk = BRW_TILEWALK_YMAJOR;
- break;
+ return GEN7_SURFACE_TILING_Y;
+ default:
+ return GEN7_SURFACE_TILING_NONE;
}
}
-void
-gen7_set_surface_msaa(struct gen7_surface_state *surf, unsigned num_samples,
- enum intel_msaa_layout layout)
+uint32_t
+gen7_surface_msaa_bits(unsigned num_samples, enum intel_msaa_layout layout)
{
+ uint32_t ss4 = 0;
+
if (num_samples > 4)
- surf->ss4.num_multisamples = GEN7_SURFACE_MULTISAMPLECOUNT_8;
+ ss4 |= GEN7_SURFACE_MULTISAMPLECOUNT_8;
else if (num_samples > 1)
- surf->ss4.num_multisamples = GEN7_SURFACE_MULTISAMPLECOUNT_4;
+ ss4 |= GEN7_SURFACE_MULTISAMPLECOUNT_4;
else
- surf->ss4.num_multisamples = GEN7_SURFACE_MULTISAMPLECOUNT_1;
+ ss4 |= GEN7_SURFACE_MULTISAMPLECOUNT_1;
- surf->ss4.multisampled_surface_storage_format =
- layout == INTEL_MSAA_LAYOUT_IMS ?
- GEN7_SURFACE_MSFMT_DEPTH_STENCIL :
- GEN7_SURFACE_MSFMT_MSS;
+ if (layout == INTEL_MSAA_LAYOUT_IMS)
+ ss4 |= GEN7_SURFACE_MSFMT_DEPTH_STENCIL;
+ else
+ ss4 |= GEN7_SURFACE_MSFMT_MSS;
+
+ return ss4;
}
void
gen7_set_surface_mcs_info(struct brw_context *brw,
- struct gen7_surface_state *surf,
+ uint32_t *surf,
uint32_t surf_offset,
const struct intel_mipmap_tree *mcs_mt,
bool is_render_target)
/* Compute the pitch in units of tiles. To do this we need to divide the
* pitch in bytes by 128, since a single Y-tile is 128 bytes wide.
*/
- unsigned pitch_bytes = mcs_mt->region->pitch * mcs_mt->cpp;
- unsigned pitch_tiles = pitch_bytes / 128;
+ unsigned pitch_tiles = mcs_mt->region->pitch / 128;
/* The upper 20 bits of surface state DWORD 6 are the upper 20 bits of the
* GPU address of the MCS buffer; the lower 12 bits contain other control
* the necessary address translation.
*/
assert ((mcs_mt->region->bo->offset & 0xfff) == 0);
- surf->ss6.mcs_enabled.mcs_enable = 1;
- surf->ss6.mcs_enabled.mcs_surface_pitch = pitch_tiles - 1;
- surf->ss6.mcs_enabled.mcs_base_address = mcs_mt->region->bo->offset >> 12;
- drm_intel_bo_emit_reloc(brw->intel.batch.bo,
- surf_offset +
- offsetof(struct gen7_surface_state, ss6),
+
+ surf[6] = GEN7_SURFACE_MCS_ENABLE |
+ SET_FIELD(pitch_tiles - 1, GEN7_SURFACE_MCS_PITCH) |
+ mcs_mt->region->bo->offset;
+
+ drm_intel_bo_emit_reloc(brw->batch.bo,
+ surf_offset + 6 * 4,
mcs_mt->region->bo,
- surf->ss6.raw_data & 0xfff,
+ surf[6] & 0xfff,
is_render_target ? I915_GEM_DOMAIN_RENDER
: I915_GEM_DOMAIN_SAMPLER,
is_render_target ? I915_GEM_DOMAIN_RENDER : 0);
void
-gen7_check_surface_setup(struct gen7_surface_state *surf,
- bool is_render_target)
+gen7_check_surface_setup(uint32_t *surf, bool is_render_target)
{
- bool is_multisampled =
- surf->ss4.num_multisamples != GEN7_SURFACE_MULTISAMPLECOUNT_1;
- /* From the Graphics BSpec: vol5c Shared Functions [SNB+] > State >
- * SURFACE_STATE > SURFACE_STATE for most messages [DevIVB]: Surface Array
- * Spacing:
+ unsigned num_multisamples = surf[4] & INTEL_MASK(5, 3);
+ unsigned multisampled_surface_storage_format = surf[4] & (1 << 6);
+ unsigned surface_array_spacing = surf[0] & (1 << 10);
+ bool is_multisampled = num_multisamples != GEN7_SURFACE_MULTISAMPLECOUNT_1;
+
+ (void) surface_array_spacing;
+
+ /* From the Ivybridge PRM, Volume 4 Part 1, page 66 (RENDER_SURFACE_STATE
+ * dword 0 bit 10 "Surface Array Spacing" Programming Notes):
*
* If Multisampled Surface Storage Format is MSFMT_MSS and Number of
* Multisamples is not MULTISAMPLECOUNT_1, this field must be set to
* ARYSPC_LOD0.
*/
- if (surf->ss4.multisampled_surface_storage_format == GEN7_SURFACE_MSFMT_MSS
+ if (multisampled_surface_storage_format == GEN7_SURFACE_MSFMT_MSS
&& is_multisampled)
- assert(surf->ss0.surface_array_spacing == GEN7_SURFACE_ARYSPC_LOD0);
+ assert(surface_array_spacing == GEN7_SURFACE_ARYSPC_LOD0);
- /* From the Graphics BSpec: vol5c Shared Functions [SNB+] > State >
- * SURFACE_STATE > SURFACE_STATE for most messages [DevIVB]: Multisampled
- * Surface Storage Format:
+ /* From the Ivybridge PRM, Volume 4 Part 1, page 72 (RENDER_SURFACE_STATE
+ * dword 4 bit 6 "Multisampled Surface Storage" Programming Notes):
*
* All multisampled render target surfaces must have this field set to
* MSFMT_MSS.
* This field is ignored if Number of Multisamples is MULTISAMPLECOUNT_1.
*/
if (is_render_target && is_multisampled) {
- assert(surf->ss4.multisampled_surface_storage_format ==
- GEN7_SURFACE_MSFMT_MSS);
+ assert(multisampled_surface_storage_format == GEN7_SURFACE_MSFMT_MSS);
}
- /* From the Graphics BSpec: vol5c Shared Functions [SNB+] > State >
- * SURFACE_STATE > SURFACE_STATE for most messages [DevIVB]: Multisampled
- * Surface Storage Format:
+ /* From the Ivybridge PRM, Volume 4 Part 1, page 72 (RENDER_SURFACE_STATE
+ * dword 4 bit 6 "Multisampled Surface Storage Format" Errata):
*
* If the surface’s Number of Multisamples is MULTISAMPLECOUNT_8, Width
* is >= 8192 (meaning the actual surface width is >= 8193 pixels), this
* field must be set to MSFMT_MSS.
*/
- if (surf->ss4.num_multisamples == GEN7_SURFACE_MULTISAMPLECOUNT_8 &&
- surf->ss2.width >= 8192) {
- assert(surf->ss4.multisampled_surface_storage_format ==
- GEN7_SURFACE_MSFMT_MSS);
+ uint32_t width = GET_FIELD(surf[2], GEN7_SURFACE_WIDTH) + 1;
+ if (num_multisamples == GEN7_SURFACE_MULTISAMPLECOUNT_8 && width >= 8193) {
+ assert(multisampled_surface_storage_format == GEN7_SURFACE_MSFMT_MSS);
}
- /* From the Graphics BSpec: vol5c Shared Functions [SNB+] > State >
- * SURFACE_STATE > SURFACE_STATE for most messages [DevIVB]: Multisampled
- * Surface Storage Format:
+ /* From the Ivybridge PRM, Volume 4 Part 1, page 72 (RENDER_SURFACE_STATE
+ * dword 4 bit 6 "Multisampled Surface Storage Format" Errata):
*
* If the surface’s Number of Multisamples is MULTISAMPLECOUNT_8,
* ((Depth+1) * (Height+1)) is > 4,194,304, OR if the surface’s Number of
* following: I24X8_UNORM, L24X8_UNORM, A24X8_UNORM, or
* R24_UNORM_X8_TYPELESS.
*
- * But also:
+ * But also (from the Programming Notes):
*
* This field is ignored if Number of Multisamples is MULTISAMPLECOUNT_1.
*/
- uint32_t depth = surf->ss3.depth + 1;
- uint32_t height = surf->ss2.height + 1;
- if (surf->ss4.num_multisamples == GEN7_SURFACE_MULTISAMPLECOUNT_8 &&
+ uint32_t depth = GET_FIELD(surf[3], BRW_SURFACE_DEPTH) + 1;
+ uint32_t height = GET_FIELD(surf[2], GEN7_SURFACE_HEIGHT) + 1;
+ if (num_multisamples == GEN7_SURFACE_MULTISAMPLECOUNT_8 &&
depth * height > 4194304) {
- assert(surf->ss4.multisampled_surface_storage_format ==
+ assert(multisampled_surface_storage_format ==
GEN7_SURFACE_MSFMT_DEPTH_STENCIL);
}
- if (surf->ss4.num_multisamples == GEN7_SURFACE_MULTISAMPLECOUNT_4 &&
+ if (num_multisamples == GEN7_SURFACE_MULTISAMPLECOUNT_4 &&
depth * height > 8388608) {
- assert(surf->ss4.multisampled_surface_storage_format ==
+ assert(multisampled_surface_storage_format ==
GEN7_SURFACE_MSFMT_DEPTH_STENCIL);
}
if (is_multisampled) {
- switch (surf->ss0.surface_format) {
+ switch (GET_FIELD(surf[0], BRW_SURFACE_FORMAT)) {
case BRW_SURFACEFORMAT_I24X8_UNORM:
case BRW_SURFACEFORMAT_L24X8_UNORM:
case BRW_SURFACEFORMAT_A24X8_UNORM:
case BRW_SURFACEFORMAT_R24_UNORM_X8_TYPELESS:
- assert(surf->ss4.multisampled_surface_storage_format ==
+ assert(multisampled_surface_storage_format ==
GEN7_SURFACE_MSFMT_DEPTH_STENCIL);
}
}
{
struct brw_context *brw = brw_context(ctx);
struct gl_texture_object *tObj = ctx->Texture.Unit[unit]._Current;
- struct gen7_surface_state *surf;
struct intel_buffer_object *intel_obj =
intel_buffer_object(tObj->BufferObject);
drm_intel_bo *bo = intel_obj ? intel_obj->buffer : NULL;
gl_format format = tObj->_BufferObjectFormat;
- int texel_size = _mesa_get_format_bytes(format);
-
- surf = brw_state_batch(brw, AUB_TRACE_SURFACE_STATE,
- sizeof(*surf), 32, &binding_table[surf_index]);
- memset(surf, 0, sizeof(*surf));
- surf->ss0.surface_type = BRW_SURFACE_BUFFER;
- surf->ss0.surface_format = brw_format_for_mesa_format(format);
+ uint32_t *surf = brw_state_batch(brw, AUB_TRACE_SURFACE_STATE,
+ 8 * 4, 32, &binding_table[surf_index]);
+ memset(surf, 0, 8 * 4);
- surf->ss0.render_cache_read_write = 1;
-
- if (surf->ss0.surface_format == 0 && format != MESA_FORMAT_RGBA_FLOAT32) {
+ uint32_t surface_format = brw_format_for_mesa_format(format);
+ if (surface_format == 0 && format != MESA_FORMAT_RGBA_FLOAT32) {
_mesa_problem(NULL, "bad format %s for texture buffer\n",
- _mesa_get_format_name(format));
+ _mesa_get_format_name(format));
}
+ surf[0] = BRW_SURFACE_BUFFER << BRW_SURFACE_TYPE_SHIFT |
+ surface_format << BRW_SURFACE_FORMAT_SHIFT |
+ BRW_SURFACE_RC_READ_WRITE;
+
if (bo) {
- surf->ss1.base_addr = bo->offset; /* reloc */
+ surf[1] = bo->offset; /* reloc */
- /* Emit relocation to surface contents. Section 5.1.1 of the gen4
- * bspec ("Data Cache") says that the data cache does not exist as
- * a separate cache and is just the sampler cache.
- */
- drm_intel_bo_emit_reloc(brw->intel.batch.bo,
- (binding_table[surf_index] +
- offsetof(struct gen7_surface_state, ss1)),
+ drm_intel_bo_emit_reloc(brw->batch.bo,
+ binding_table[surf_index] + 4,
bo, 0,
I915_GEM_DOMAIN_SAMPLER, 0);
+ int texel_size = _mesa_get_format_bytes(format);
int w = intel_obj->Base.Size / texel_size;
- surf->ss2.width = w & 0x7f; /* bits 6:0 of size or width */
- surf->ss2.height = (w >> 7) & 0x1fff; /* bits 19:7 of size or width */
- surf->ss3.depth = (w >> 20) & 0x7f; /* bits 26:20 of size or width */
- surf->ss3.pitch = texel_size - 1;
-} else {
- surf->ss1.base_addr = 0;
- surf->ss2.width = 0;
- surf->ss2.height = 0;
- surf->ss3.depth = 0;
- surf->ss3.pitch = 0;
- }
- gen7_set_surface_tiling(surf, I915_TILING_NONE);
+ /* note that these differ from GEN6 */
+ surf[2] = SET_FIELD(w & 0x7f, GEN7_SURFACE_WIDTH) | /* bits 6:0 of size */
+ SET_FIELD((w >> 7) & 0x3fff, GEN7_SURFACE_HEIGHT); /* 20:7 */
+ surf[3] = SET_FIELD((w >> 21) & 0x3f, BRW_SURFACE_DEPTH) | /* bits 26:21 */
+ (texel_size - 1);
+ }
gen7_check_surface_setup(surf, false /* is_render_target */);
}
struct intel_mipmap_tree *mt = intelObj->mt;
struct gl_texture_image *firstImage = tObj->Image[0][tObj->BaseLevel];
struct gl_sampler_object *sampler = _mesa_get_samplerobj(ctx, unit);
- struct gen7_surface_state *surf;
- int width, height, depth;
+ uint32_t tile_x, tile_y;
if (tObj->Target == GL_TEXTURE_BUFFER) {
gen7_update_buffer_texture_surface(ctx, unit, binding_table, surf_index);
return;
}
- /* We don't support MSAA for textures. */
- assert(!mt->array_spacing_lod0);
- assert(mt->num_samples <= 1);
+ uint32_t *surf = brw_state_batch(brw, AUB_TRACE_SURFACE_STATE,
+ 8 * 4, 32, &binding_table[surf_index]);
+ memset(surf, 0, 8 * 4);
- intel_miptree_get_dimensions_for_image(firstImage, &width, &height, &depth);
+ uint32_t tex_format = translate_tex_format(brw,
+ mt->format,
+ tObj->DepthMode,
+ sampler->sRGBDecode);
- surf = brw_state_batch(brw, AUB_TRACE_SURFACE_STATE,
- sizeof(*surf), 32, &binding_table[surf_index]);
- memset(surf, 0, sizeof(*surf));
+ surf[0] = translate_tex_target(tObj->Target) << BRW_SURFACE_TYPE_SHIFT |
+ tex_format << BRW_SURFACE_FORMAT_SHIFT |
+ gen7_surface_tiling_mode(mt->region->tiling) |
+ BRW_SURFACE_CUBEFACE_ENABLES;
if (mt->align_h == 4)
- surf->ss0.vertical_alignment = 1;
+ surf[0] |= GEN7_SURFACE_VALIGN_4;
if (mt->align_w == 8)
- surf->ss0.horizontal_alignment = 1;
-
- surf->ss0.surface_type = translate_tex_target(tObj->Target);
- surf->ss0.surface_format = translate_tex_format(mt->format,
- firstImage->InternalFormat,
- tObj->DepthMode,
- sampler->sRGBDecode);
- if (tObj->Target == GL_TEXTURE_CUBE_MAP) {
- surf->ss0.cube_pos_x = 1;
- surf->ss0.cube_pos_y = 1;
- surf->ss0.cube_pos_z = 1;
- surf->ss0.cube_neg_x = 1;
- surf->ss0.cube_neg_y = 1;
- surf->ss0.cube_neg_z = 1;
- }
+ surf[0] |= GEN7_SURFACE_HALIGN_8;
- surf->ss0.is_array = depth > 1 && tObj->Target != GL_TEXTURE_3D;
+ if (mt->logical_depth0 > 1 && tObj->Target != GL_TEXTURE_3D)
+ surf[0] |= GEN7_SURFACE_IS_ARRAY;
- gen7_set_surface_tiling(surf, intelObj->mt->region->tiling);
-
- /* ss0 remaining fields:
- * - vert_line_stride (exists on gen6 but we ignore it)
- * - vert_line_stride_ofs (exists on gen6 but we ignore it)
- * - surface_array_spacing
- * - render_cache_read_write (exists on gen6 but ignored here)
- */
+ if (mt->array_spacing_lod0)
+ surf[0] |= GEN7_SURFACE_ARYSPC_LOD0;
- surf->ss1.base_addr =
- intelObj->mt->region->bo->offset + intelObj->mt->offset; /* reloc */
+ surf[1] = mt->region->bo->offset + mt->offset; /* reloc */
+ surf[1] += intel_miptree_get_tile_offsets(intelObj->mt, firstImage->Level, 0,
+ &tile_x, &tile_y);
- surf->ss2.width = width - 1;
- surf->ss2.height = height - 1;
+ surf[2] = SET_FIELD(mt->logical_width0 - 1, GEN7_SURFACE_WIDTH) |
+ SET_FIELD(mt->logical_height0 - 1, GEN7_SURFACE_HEIGHT);
+ surf[3] = SET_FIELD(mt->logical_depth0 - 1, BRW_SURFACE_DEPTH) |
+ ((intelObj->mt->region->pitch) - 1);
- surf->ss3.pitch = (intelObj->mt->region->pitch * intelObj->mt->cpp) - 1;
- surf->ss3.depth = depth - 1;
+ surf[4] = gen7_surface_msaa_bits(mt->num_samples, mt->msaa_layout);
- /* ss4: ignored? */
-
- surf->ss5.mip_count = intelObj->_MaxLevel - tObj->BaseLevel;
- surf->ss5.min_lod = 0;
-
- /* ss5 remaining fields:
- * - x_offset (N/A for textures?)
- * - y_offset (ditto)
- * - cache_control
+ assert(brw->has_surface_tile_offset || (tile_x == 0 && tile_y == 0));
+ /* Note that the low bits of these fields are missing, so
+ * there's the possibility of getting in trouble.
*/
+ surf[5] = ((tile_x / 4) << BRW_SURFACE_X_OFFSET_SHIFT |
+ (tile_y / 2) << BRW_SURFACE_Y_OFFSET_SHIFT |
+ SET_FIELD(GEN7_MOCS_L3, GEN7_SURFACE_MOCS) |
+ /* mip count */
+ (intelObj->_MaxLevel - tObj->BaseLevel));
- if (brw->intel.is_haswell) {
+ if (brw->is_haswell) {
/* Handling GL_ALPHA as a surface format override breaks 1.30+ style
* texturing functions that return a float, as our code generation always
* selects the .x channel (which would always be 0).
(firstImage->_BaseFormat == GL_DEPTH_COMPONENT ||
firstImage->_BaseFormat == GL_DEPTH_STENCIL);
- const int swizzle =
- unlikely(alpha_depth) ? SWIZZLE_XYZW : brw_get_texture_swizzle(tObj);
+ const int swizzle = unlikely(alpha_depth)
+ ? SWIZZLE_XYZW : brw_get_texture_swizzle(ctx, tObj);
- surf->ss7.shader_channel_select_r = swizzle_to_scs(GET_SWZ(swizzle, 0));
- surf->ss7.shader_channel_select_g = swizzle_to_scs(GET_SWZ(swizzle, 1));
- surf->ss7.shader_channel_select_b = swizzle_to_scs(GET_SWZ(swizzle, 2));
- surf->ss7.shader_channel_select_a = swizzle_to_scs(GET_SWZ(swizzle, 3));
+ surf[7] =
+ SET_FIELD(swizzle_to_scs(GET_SWZ(swizzle, 0)), GEN7_SURFACE_SCS_R) |
+ SET_FIELD(swizzle_to_scs(GET_SWZ(swizzle, 1)), GEN7_SURFACE_SCS_G) |
+ SET_FIELD(swizzle_to_scs(GET_SWZ(swizzle, 2)), GEN7_SURFACE_SCS_B) |
+ SET_FIELD(swizzle_to_scs(GET_SWZ(swizzle, 3)), GEN7_SURFACE_SCS_A);
}
/* Emit relocation to surface contents */
- drm_intel_bo_emit_reloc(brw->intel.batch.bo,
- binding_table[surf_index] +
- offsetof(struct gen7_surface_state, ss1),
- intelObj->mt->region->bo, intelObj->mt->offset,
+ drm_intel_bo_emit_reloc(brw->batch.bo,
+ binding_table[surf_index] + 4,
+ intelObj->mt->region->bo,
+ surf[1] - intelObj->mt->region->bo->offset,
I915_GEM_DOMAIN_SAMPLER, 0);
gen7_check_surface_setup(surf, false /* is_render_target */);
gen7_create_constant_surface(struct brw_context *brw,
drm_intel_bo *bo,
uint32_t offset,
- int width,
- uint32_t *out_offset)
+ uint32_t size,
+ uint32_t *out_offset,
+ bool dword_pitch)
{
- const GLint w = width - 1;
- struct gen7_surface_state *surf;
+ uint32_t stride = dword_pitch ? 4 : 16;
+ uint32_t elements = ALIGN(size, stride) / stride;
+ const GLint w = elements - 1;
- surf = brw_state_batch(brw, AUB_TRACE_SURFACE_STATE,
- sizeof(*surf), 32, out_offset);
- memset(surf, 0, sizeof(*surf));
+ uint32_t *surf = brw_state_batch(brw, AUB_TRACE_SURFACE_STATE,
+ 8 * 4, 32, out_offset);
+ memset(surf, 0, 8 * 4);
- surf->ss0.surface_type = BRW_SURFACE_BUFFER;
- surf->ss0.surface_format = BRW_SURFACEFORMAT_R32G32B32A32_FLOAT;
-
- surf->ss0.render_cache_read_write = 1;
+ surf[0] = BRW_SURFACE_BUFFER << BRW_SURFACE_TYPE_SHIFT |
+ BRW_SURFACEFORMAT_R32G32B32A32_FLOAT << BRW_SURFACE_FORMAT_SHIFT |
+ BRW_SURFACE_RC_READ_WRITE;
assert(bo);
- surf->ss1.base_addr = bo->offset + offset; /* reloc */
-
- surf->ss2.width = w & 0x7f; /* bits 6:0 of size or width */
- surf->ss2.height = (w >> 7) & 0x1fff; /* bits 19:7 of size or width */
- surf->ss3.depth = (w >> 20) & 0x7f; /* bits 26:20 of size or width */
- surf->ss3.pitch = (16 - 1); /* stride between samples */
- gen7_set_surface_tiling(surf, I915_TILING_NONE); /* tiling now allowed */
-
- if (brw->intel.is_haswell) {
- surf->ss7.shader_channel_select_r = HSW_SCS_RED;
- surf->ss7.shader_channel_select_g = HSW_SCS_GREEN;
- surf->ss7.shader_channel_select_b = HSW_SCS_BLUE;
- surf->ss7.shader_channel_select_a = HSW_SCS_ALPHA;
+ surf[1] = bo->offset + offset; /* reloc */
+
+ /* note that these differ from GEN6 */
+ surf[2] = SET_FIELD(w & 0x7f, GEN7_SURFACE_WIDTH) |
+ SET_FIELD((w >> 7) & 0x3fff, GEN7_SURFACE_HEIGHT);
+ surf[3] = SET_FIELD((w >> 21) & 0x3f, BRW_SURFACE_DEPTH) |
+ (stride - 1);
+
+ if (brw->is_haswell) {
+ surf[7] = SET_FIELD(HSW_SCS_RED, GEN7_SURFACE_SCS_R) |
+ SET_FIELD(HSW_SCS_GREEN, GEN7_SURFACE_SCS_G) |
+ SET_FIELD(HSW_SCS_BLUE, GEN7_SURFACE_SCS_B) |
+ SET_FIELD(HSW_SCS_ALPHA, GEN7_SURFACE_SCS_A);
}
- /* Emit relocation to surface contents. Section 5.1.1 of the gen4
- * bspec ("Data Cache") says that the data cache does not exist as
- * a separate cache and is just the sampler cache.
- */
- drm_intel_bo_emit_reloc(brw->intel.batch.bo,
- (*out_offset +
- offsetof(struct gen7_surface_state, ss1)),
+ drm_intel_bo_emit_reloc(brw->batch.bo,
+ *out_offset + 4,
bo, offset,
I915_GEM_DOMAIN_SAMPLER, 0);
gen7_check_surface_setup(surf, false /* is_render_target */);
}
+/**
+ * Create a surface for shader time.
+ */
+void
+gen7_create_shader_time_surface(struct brw_context *brw, uint32_t *out_offset)
+{
+ const int w = brw->shader_time.bo->size - 1;
+
+ uint32_t *surf = brw_state_batch(brw, AUB_TRACE_SURFACE_STATE,
+ 8 * 4, 32, out_offset);
+ memset(surf, 0, 8 * 4);
+
+ surf[0] = BRW_SURFACE_BUFFER << BRW_SURFACE_TYPE_SHIFT |
+ BRW_SURFACEFORMAT_RAW << BRW_SURFACE_FORMAT_SHIFT |
+ BRW_SURFACE_RC_READ_WRITE;
+
+ surf[1] = brw->shader_time.bo->offset; /* reloc */
+
+ /* note that these differ from GEN6 */
+ surf[2] = SET_FIELD(w & 0x7f, GEN7_SURFACE_WIDTH) |
+ SET_FIELD((w >> 7) & 0x3fff, GEN7_SURFACE_HEIGHT);
+ surf[3] = SET_FIELD((w >> 21) & 0x3f, BRW_SURFACE_DEPTH);
+
+ /* Unlike texture or renderbuffer surfaces, we only do untyped operations
+ * on the shader_time surface, so there's no need to set HSW channel
+ * overrides.
+ */
+
+ drm_intel_bo_emit_reloc(brw->batch.bo,
+ *out_offset + 4,
+ brw->shader_time.bo, 0,
+ I915_GEM_DOMAIN_SAMPLER, 0);
+
+ gen7_check_surface_setup(surf, false /* is_render_target */);
+}
+
static void
gen7_update_null_renderbuffer_surface(struct brw_context *brw, unsigned unit)
{
* depth buffer’s corresponding state for all render target surfaces,
* including null.
*/
- struct intel_context *intel = &brw->intel;
- struct gl_context *ctx = &intel->ctx;
- struct gen7_surface_state *surf;
+ struct gl_context *ctx = &brw->ctx;
/* _NEW_BUFFERS */
const struct gl_framebuffer *fb = ctx->DrawBuffer;
- surf = brw_state_batch(brw, AUB_TRACE_SURFACE_STATE,
- sizeof(*surf), 32, &brw->wm.surf_offset[unit]);
- memset(surf, 0, sizeof(*surf));
-
- surf->ss0.surface_type = BRW_SURFACE_NULL;
- surf->ss0.surface_format = BRW_SURFACEFORMAT_B8G8R8A8_UNORM;
-
- surf->ss2.width = fb->Width - 1;
- surf->ss2.height = fb->Height - 1;
+ uint32_t *surf = brw_state_batch(brw, AUB_TRACE_SURFACE_STATE, 8 * 4, 32,
+ &brw->wm.surf_offset[SURF_INDEX_DRAW(unit)]);
+ memset(surf, 0, 8 * 4);
- /* From the Ivy bridge PRM, Vol4 Part1 p65 (Tiled Surface: Programming Notes):
- *
- * If Surface Type is SURFTYPE_NULL, this field must be TRUE.
+ /* From the Ivybridge PRM, Volume 4, Part 1, page 65,
+ * Tiled Surface: Programming Notes:
+ * "If Surface Type is SURFTYPE_NULL, this field must be TRUE."
*/
- gen7_set_surface_tiling(surf, I915_TILING_Y);
+ surf[0] = BRW_SURFACE_NULL << BRW_SURFACE_TYPE_SHIFT |
+ BRW_SURFACEFORMAT_B8G8R8A8_UNORM << BRW_SURFACE_FORMAT_SHIFT |
+ GEN7_SURFACE_TILING_Y;
+
+ surf[2] = SET_FIELD(fb->Width - 1, GEN7_SURFACE_WIDTH) |
+ SET_FIELD(fb->Height - 1, GEN7_SURFACE_HEIGHT);
gen7_check_surface_setup(surf, true /* is_render_target */);
}
static void
gen7_update_renderbuffer_surface(struct brw_context *brw,
struct gl_renderbuffer *rb,
+ bool layered,
unsigned int unit)
{
- struct intel_context *intel = &brw->intel;
- struct gl_context *ctx = &intel->ctx;
+ struct gl_context *ctx = &brw->ctx;
struct intel_renderbuffer *irb = intel_renderbuffer(rb);
struct intel_region *region = irb->mt->region;
- struct gen7_surface_state *surf;
- uint32_t tile_x, tile_y;
- gl_format rb_format = intel_rb_format(irb);
+ uint32_t format;
+ /* _NEW_BUFFERS */
+ gl_format rb_format = _mesa_get_render_format(ctx, intel_rb_format(irb));
+ uint32_t surftype;
+ bool is_array = false;
+ int depth = MAX2(rb->Depth, 1);
+ int min_array_element;
+ const uint8_t mocs = GEN7_MOCS_L3;
+ GLenum gl_target = rb->TexImage ?
+ rb->TexImage->TexObject->Target : GL_TEXTURE_2D;
+
+ uint32_t surf_index = SURF_INDEX_DRAW(unit);
- surf = brw_state_batch(brw, AUB_TRACE_SURFACE_STATE,
- sizeof(*surf), 32, &brw->wm.surf_offset[unit]);
- memset(surf, 0, sizeof(*surf));
+ uint32_t *surf = brw_state_batch(brw, AUB_TRACE_SURFACE_STATE, 8 * 4, 32,
+ &brw->wm.surf_offset[surf_index]);
+ memset(surf, 0, 8 * 4);
+
+ intel_miptree_used_for_rendering(irb->mt);
/* Render targets can't use IMS layout */
assert(irb->mt->msaa_layout != INTEL_MSAA_LAYOUT_IMS);
- if (irb->mt->align_h == 4)
- surf->ss0.vertical_alignment = 1;
- if (irb->mt->align_w == 8)
- surf->ss0.horizontal_alignment = 1;
-
- switch (rb_format) {
- case MESA_FORMAT_SARGB8:
- /* _NEW_BUFFERS
- *
- * Without GL_EXT_framebuffer_sRGB we shouldn't bind sRGB surfaces to the
- * blend/update as sRGB.
- */
- if (ctx->Color.sRGBEnabled)
- surf->ss0.surface_format = brw_format_for_mesa_format(rb_format);
- else
- surf->ss0.surface_format = BRW_SURFACEFORMAT_B8G8R8A8_UNORM;
+ assert(brw_render_target_supported(brw, rb));
+ format = brw->render_target_format[rb_format];
+ if (unlikely(!brw->format_supported_as_render_target[rb_format])) {
+ _mesa_problem(ctx, "%s: renderbuffer format %s unsupported\n",
+ __FUNCTION__, _mesa_get_format_name(rb_format));
+ }
+
+ switch (gl_target) {
+ case GL_TEXTURE_CUBE_MAP_ARRAY:
+ case GL_TEXTURE_CUBE_MAP:
+ surftype = BRW_SURFACE_2D;
+ is_array = true;
+ depth *= 6;
break;
default:
- assert(brw_render_target_supported(intel, rb));
- surf->ss0.surface_format = brw->render_target_format[rb_format];
- if (unlikely(!brw->format_supported_as_render_target[rb_format])) {
- _mesa_problem(ctx, "%s: renderbuffer format %s unsupported\n",
- __FUNCTION__, _mesa_get_format_name(rb_format));
- }
- break;
+ surftype = translate_tex_target(gl_target);
+ is_array = _mesa_tex_target_is_array(gl_target);
+ break;
}
- surf->ss0.surface_type = BRW_SURFACE_2D;
- surf->ss0.surface_array_spacing = irb->mt->array_spacing_lod0 ?
- GEN7_SURFACE_ARYSPC_LOD0 : GEN7_SURFACE_ARYSPC_FULL;
+ if (layered) {
+ min_array_element = 0;
+ } else if (irb->mt->num_samples > 1) {
+ min_array_element = irb->mt_layer / irb->mt->num_samples;
+ } else {
+ min_array_element = irb->mt_layer;
+ }
+
+ surf[0] = surftype << BRW_SURFACE_TYPE_SHIFT |
+ format << BRW_SURFACE_FORMAT_SHIFT |
+ (irb->mt->array_spacing_lod0 ? GEN7_SURFACE_ARYSPC_LOD0
+ : GEN7_SURFACE_ARYSPC_FULL) |
+ gen7_surface_tiling_mode(region->tiling);
+
+ if (irb->mt->align_h == 4)
+ surf[0] |= GEN7_SURFACE_VALIGN_4;
+ if (irb->mt->align_w == 8)
+ surf[0] |= GEN7_SURFACE_HALIGN_8;
+
+ if (is_array) {
+ surf[0] |= GEN7_SURFACE_IS_ARRAY;
+ }
- /* reloc */
- surf->ss1.base_addr = intel_renderbuffer_tile_offsets(irb, &tile_x, &tile_y);
- surf->ss1.base_addr += region->bo->offset; /* reloc */
+ surf[1] = region->bo->offset;
assert(brw->has_surface_tile_offset);
- /* Note that the low bits of these fields are missing, so
- * there's the possibility of getting in trouble.
- */
- assert(tile_x % 4 == 0);
- assert(tile_y % 2 == 0);
- surf->ss5.x_offset = tile_x / 4;
- surf->ss5.y_offset = tile_y / 2;
- surf->ss2.width = rb->Width - 1;
- surf->ss2.height = rb->Height - 1;
- gen7_set_surface_tiling(surf, region->tiling);
- surf->ss3.pitch = (region->pitch * region->cpp) - 1;
+ surf[5] = SET_FIELD(mocs, GEN7_SURFACE_MOCS) |
+ (irb->mt_level - irb->mt->first_level);
- gen7_set_surface_msaa(surf, irb->mt->num_samples, irb->mt->msaa_layout);
+ surf[2] = SET_FIELD(irb->mt->logical_width0 - 1, GEN7_SURFACE_WIDTH) |
+ SET_FIELD(irb->mt->logical_height0 - 1, GEN7_SURFACE_HEIGHT);
- if (irb->mt->msaa_layout == INTEL_MSAA_LAYOUT_CMS) {
- gen7_set_surface_mcs_info(brw, surf, brw->wm.surf_offset[unit],
- irb->mt->mcs_mt, true /* is_render_target */);
+ surf[3] = ((depth - 1) << BRW_SURFACE_DEPTH_SHIFT) |
+ (region->pitch - 1);
+
+ surf[4] = gen7_surface_msaa_bits(irb->mt->num_samples, irb->mt->msaa_layout) |
+ min_array_element << GEN7_SURFACE_MIN_ARRAY_ELEMENT_SHIFT |
+ (depth - 1) << GEN7_SURFACE_RENDER_TARGET_VIEW_EXTENT_SHIFT;
+
+ if (irb->mt->mcs_mt) {
+ gen7_set_surface_mcs_info(brw, surf, brw->wm.surf_offset[surf_index],
+ irb->mt->mcs_mt, true /* is RT */);
}
- if (intel->is_haswell) {
- surf->ss7.shader_channel_select_r = HSW_SCS_RED;
- surf->ss7.shader_channel_select_g = HSW_SCS_GREEN;
- surf->ss7.shader_channel_select_b = HSW_SCS_BLUE;
- surf->ss7.shader_channel_select_a = HSW_SCS_ALPHA;
+ surf[7] = irb->mt->fast_clear_color_value;
+
+ if (brw->is_haswell) {
+ surf[7] |= (SET_FIELD(HSW_SCS_RED, GEN7_SURFACE_SCS_R) |
+ SET_FIELD(HSW_SCS_GREEN, GEN7_SURFACE_SCS_G) |
+ SET_FIELD(HSW_SCS_BLUE, GEN7_SURFACE_SCS_B) |
+ SET_FIELD(HSW_SCS_ALPHA, GEN7_SURFACE_SCS_A));
}
- drm_intel_bo_emit_reloc(brw->intel.batch.bo,
- brw->wm.surf_offset[unit] +
- offsetof(struct gen7_surface_state, ss1),
+ drm_intel_bo_emit_reloc(brw->batch.bo,
+ brw->wm.surf_offset[surf_index] + 4,
region->bo,
- surf->ss1.base_addr - region->bo->offset,
+ surf[1] - region->bo->offset,
I915_GEM_DOMAIN_RENDER,
I915_GEM_DOMAIN_RENDER);
void
gen7_init_vtable_surface_functions(struct brw_context *brw)
{
- struct intel_context *intel = &brw->intel;
-
- intel->vtbl.update_texture_surface = gen7_update_texture_surface;
- intel->vtbl.update_renderbuffer_surface = gen7_update_renderbuffer_surface;
- intel->vtbl.update_null_renderbuffer_surface =
+ brw->vtbl.update_texture_surface = gen7_update_texture_surface;
+ brw->vtbl.update_renderbuffer_surface = gen7_update_renderbuffer_surface;
+ brw->vtbl.update_null_renderbuffer_surface =
gen7_update_null_renderbuffer_surface;
- intel->vtbl.create_constant_surface = gen7_create_constant_surface;
+ brw->vtbl.create_constant_surface = gen7_create_constant_surface;
}