#define GEN7_SURFACE_MIN_LOD_SHIFT 4
#define GEN7_SURFACE_MIN_LOD_MASK INTEL_MASK(7, 4)
+#define GEN7_SURFACE_MOCS_SHIFT 16
+#define GEN7_SURFACE_MOCS_MASK INTEL_MASK(19, 16)
+
/* Surface state DW6 */
#define GEN7_SURFACE_MCS_ENABLE (1 << 0)
#define GEN7_SURFACE_MCS_PITCH_SHIFT 3
#define GEN7_SURFACE_MCS_PITCH_MASK INTEL_MASK(11, 3)
/* Surface state DW7 */
+#define GEN7_SURFACE_CLEAR_COLOR_SHIFT 28
#define GEN7_SURFACE_SCS_R_SHIFT 25
#define GEN7_SURFACE_SCS_R_MASK INTEL_MASK(27, 25)
#define GEN7_SURFACE_SCS_G_SHIFT 22
BRW_OPCODE_XOR = 7,
BRW_OPCODE_SHR = 8,
BRW_OPCODE_SHL = 9,
- BRW_OPCODE_RSR = 10,
- BRW_OPCODE_RSL = 11,
BRW_OPCODE_ASR = 12,
BRW_OPCODE_CMP = 16,
BRW_OPCODE_CMPN = 17,
BRW_OPCODE_DPH = 85,
BRW_OPCODE_DP3 = 86,
BRW_OPCODE_DP2 = 87,
- BRW_OPCODE_DPA2 = 88,
BRW_OPCODE_LINE = 89,
BRW_OPCODE_PLN = 90,
BRW_OPCODE_MAD = 91,
VS_OPCODE_SCRATCH_WRITE,
VS_OPCODE_PULL_CONSTANT_LOAD,
VS_OPCODE_PULL_CONSTANT_LOAD_GEN7,
+ VS_OPCODE_UNPACK_FLAGS_SIMD4X2,
+
+ /**
+ * Write geometry shader output data to the URB.
+ *
+ * Unlike VS_OPCODE_URB_WRITE, this opcode doesn't do an implied move from
+ * R0 to the first MRF. This allows the geometry shader to override the
+ * "Slot {0,1} Offset" fields in the message header.
+ */
+ GS_OPCODE_URB_WRITE,
+
+ /**
+ * Terminate the geometry shader thread by doing an empty URB write.
+ *
+ * This opcode doesn't do an implied move from R0 to the first MRF. This
+ * allows the geometry shader to override the "GS Number of Output Vertices
+ * for Slot {0,1}" fields in the message header.
+ */
+ GS_OPCODE_THREAD_END,
+
+ /**
+ * Set the "Slot {0,1} Offset" fields of a URB_WRITE message header.
+ *
+ * - dst is the MRF containing the message header.
+ *
+ * - src0.x indicates which portion of the URB should be written to (e.g. a
+ * vertex number)
+ *
+ * - src1 is an immediate multiplier which will be applied to src0
+ * (e.g. the size of a single vertex in the URB).
+ *
+ * Note: the hardware will apply this offset *in addition to* the offset in
+ * vec4_instruction::offset.
+ */
+ GS_OPCODE_SET_WRITE_OFFSET,
+
+ /**
+ * Set the "GS Number of Output Vertices for Slot {0,1}" fields of a
+ * URB_WRITE message header.
+ *
+ * - dst is the MRF containing the message header.
+ *
+ * - src0.x is the vertex count. The upper 16 bits will be ignored.
+ */
+ GS_OPCODE_SET_VERTEX_COUNT,
+
+ /**
+ * Set DWORD 2 of dst to the immediate value in src. Used by geometry
+ * shaders to initialize DWORD 2 of R0, which needs to be 0 in order for
+ * scratch reads and writes to operate correctly.
+ */
+ GS_OPCODE_SET_DWORD_2_IMMED,
};
#define BRW_PREDICATE_NONE 0
* In particular, see the following tables:
* - G45 PRM, Volume 4, Table 14-15 "Message Descriptor Definition"
* - Sandybridge PRM, Volume 4 Part 2, Table 8-16 "Extended Message Descriptor"
- * - BSpec, Volume 1a (GPU Overview) / Graphics Processing Engine (GPE) /
- * Overview / GPE Function IDs
+ * - Ivybridge PRM, Volume 1 Part 1, section 3.2.7 "GPE Function IDs"
*/
enum brw_message_target {
BRW_SFID_NULL = 0,
# define GEN7_URB_ENTRY_SIZE_SHIFT 16
# define GEN7_URB_STARTING_ADDRESS_SHIFT 25
+/* "GS URB Entry Allocation Size" is a U9-1 field, so the maximum gs_size
+ * is 2^9, or 512. It's counted in multiples of 64 bytes.
+ */
+#define GEN7_MAX_GS_URB_ENTRY_SIZE_BYTES (512*64)
+
#define _3DSTATE_PUSH_CONSTANT_ALLOC_VS 0x7912 /* GEN7+ */
+#define _3DSTATE_PUSH_CONSTANT_ALLOC_GS 0x7915 /* GEN7+ */
#define _3DSTATE_PUSH_CONSTANT_ALLOC_PS 0x7916 /* GEN7+ */
# define GEN7_PUSH_CONSTANT_BUFFER_OFFSET_SHIFT 16
# define BRW_GS_EDGE_INDICATOR_0 (1 << 8)
# define BRW_GS_EDGE_INDICATOR_1 (1 << 9)
+/* 3DSTATE_GS "Output Vertex Size" has an effective maximum of 62. It's
+ * counted in multiples of 16 bytes.
+ */
+#define GEN7_MAX_GS_OUTPUT_VERTEX_SIZE_BYTES (62*16)
+
#define _3DSTATE_HS 0x781B /* GEN7+ */
#define _3DSTATE_TE 0x781C /* GEN7+ */
#define _3DSTATE_DS 0x781D /* GEN7+ */
# define GEN7_PS_PUSH_CONSTANT_ENABLE (1 << 11)
# define GEN7_PS_ATTRIBUTE_ENABLE (1 << 10)
# define GEN7_PS_OMASK_TO_RENDER_TARGET (1 << 9)
+# define GEN7_PS_RENDER_TARGET_FAST_CLEAR_ENABLE (1 << 8)
# define GEN7_PS_DUAL_SOURCE_BLEND_ENABLE (1 << 7)
+# define GEN7_PS_RENDER_TARGET_RESOLVE_ENABLE (1 << 6)
# define GEN7_PS_POSOFFSET_NONE (0 << 3)
# define GEN7_PS_POSOFFSET_CENTROID (2 << 3)
# define GEN7_PS_POSOFFSET_SAMPLE (3 << 3)
*/
#define BRW_MAX_NUM_BUFFER_ENTRIES (1 << 27)
+/* Memory Object Control State:
+ * Specifying zero for L3 means "uncached in L3", at least on Haswell
+ * and Baytrail, since there are no PTE flags for setting L3 cacheability.
+ * On Ivybridge, the PTEs do have a cache-in-L3 bit, so setting MOCS to 0
+ * may still respect that.
+ */
+#define GEN7_MOCS_L3 1
+
+/* Ivybridge only: cache in LLC.
+ * Specifying zero here means to use the PTE values set by the kernel;
+ * non-zero overrides the PTE values.
+ */
+#define IVB_MOCS_LLC (1 << 1)
+
+/* Baytrail only: snoop in CPU cache */
+#define BYT_MOCS_SNOOP (1 << 1)
+
+/* Haswell only: LLC/eLLC controls (write-back or uncached).
+ * Specifying zero here means to use the PTE values set by the kernel,
+ * which is useful since it offers additional control (write-through
+ * cacheing and age). Non-zero overrides the PTE values.
+ */
+#define HSW_MOCS_UC_LLC_UC_ELLC (1 << 1)
+#define HSW_MOCS_WB_LLC_WB_ELLC (2 << 1)
+#define HSW_MOCS_UC_LLC_WB_ELLC (3 << 1)
+
#include "intel_chipset.h"
#endif