[Mesa-dev] [PATCH v2 2/3] i965: Add a SHADER_OPCODE_URB_READ_SIMD8_PER_SLOT opcode.

[Kenneth Graunke]

We need to use per-slot offsets when there's non-uniform indexing,
as each SIMD channel could have a different index.  We want to use
them for any non-constant index (even if uniform), as it lives in
the message header instead of the descriptor, allowing us to set
offsets in GRFs rather than immediates.

Signed-off-by: Kenneth Graunke <kenneth at whitecape.org>
---
 src/mesa/drivers/dri/i965/brw_defines.h        | 7 ++-----
 src/mesa/drivers/dri/i965/brw_fs.cpp           | 2 ++
 src/mesa/drivers/dri/i965/brw_fs_generator.cpp | 4 ++++
 src/mesa/drivers/dri/i965/brw_shader.cpp       | 2 ++
 4 files changed, 10 insertions(+), 5 deletions(-)

Unchanged from v1.

diff --git a/src/mesa/drivers/dri/i965/brw_defines.h b/src/mesa/drivers/dri/i965/brw_defines.h
index 6e1cfed..176da83 100644
--- a/src/mesa/drivers/dri/i965/brw_defines.h
+++ b/src/mesa/drivers/dri/i965/brw_defines.h
@@ -1031,13 +1031,10 @@ enum opcode {
    SHADER_OPCODE_GEN7_SCRATCH_READ,
 
    /**
-    * Gen8+ SIMD8 URB Read message.
-    *
-    * Source 0: The header register, containing URB handles (g1).
-    *
-    * Currently only supports constant offsets, in inst->offset.
+    * Gen8+ SIMD8 URB Read messages.
     */
    SHADER_OPCODE_URB_READ_SIMD8,
+   SHADER_OPCODE_URB_READ_SIMD8_PER_SLOT,
 
    SHADER_OPCODE_URB_WRITE_SIMD8,
    SHADER_OPCODE_URB_WRITE_SIMD8_PER_SLOT,
diff --git a/src/mesa/drivers/dri/i965/brw_fs.cpp b/src/mesa/drivers/dri/i965/brw_fs.cpp
index bf8a4a6..9396cf2 100644
--- a/src/mesa/drivers/dri/i965/brw_fs.cpp
+++ b/src/mesa/drivers/dri/i965/brw_fs.cpp
@@ -284,6 +284,7 @@ fs_inst::is_send_from_grf() const
    case SHADER_OPCODE_URB_WRITE_SIMD8_MASKED:
    case SHADER_OPCODE_URB_WRITE_SIMD8_MASKED_PER_SLOT:
    case SHADER_OPCODE_URB_READ_SIMD8:
+   case SHADER_OPCODE_URB_READ_SIMD8_PER_SLOT:
       return true;
    case FS_OPCODE_UNIFORM_PULL_CONSTANT_LOAD:
       return src[1].file == GRF;
@@ -810,6 +811,7 @@ fs_inst::regs_read(int arg) const
    case SHADER_OPCODE_URB_WRITE_SIMD8_MASKED:
    case SHADER_OPCODE_URB_WRITE_SIMD8_MASKED_PER_SLOT:
    case SHADER_OPCODE_URB_READ_SIMD8:
+   case SHADER_OPCODE_URB_READ_SIMD8_PER_SLOT:
    case SHADER_OPCODE_UNTYPED_ATOMIC:
    case SHADER_OPCODE_UNTYPED_SURFACE_READ:
    case SHADER_OPCODE_UNTYPED_SURFACE_WRITE:
diff --git a/src/mesa/drivers/dri/i965/brw_fs_generator.cpp b/src/mesa/drivers/dri/i965/brw_fs_generator.cpp
index 97a85bb..48e350f 100644
--- a/src/mesa/drivers/dri/i965/brw_fs_generator.cpp
+++ b/src/mesa/drivers/dri/i965/brw_fs_generator.cpp
@@ -413,6 +413,9 @@ fs_generator::generate_urb_read(fs_inst *inst,
    brw_inst_set_sfid(p->devinfo, send, BRW_SFID_URB);
    brw_inst_set_urb_opcode(p->devinfo, send, GEN8_URB_OPCODE_SIMD8_READ);
 
+   if (inst->opcode == SHADER_OPCODE_URB_READ_SIMD8_PER_SLOT)
+      brw_inst_set_urb_per_slot_offset(p->devinfo, send, true);
+
    brw_inst_set_mlen(p->devinfo, send, inst->mlen);
    brw_inst_set_rlen(p->devinfo, send, inst->regs_written);
    brw_inst_set_header_present(p->devinfo, send, true);
@@ -2105,6 +2108,7 @@ fs_generator::generate_code(const cfg_t *cfg, int dispatch_width)
          break;
 
       case SHADER_OPCODE_URB_READ_SIMD8:
+      case SHADER_OPCODE_URB_READ_SIMD8_PER_SLOT:
          generate_urb_read(inst, dst, src[0]);
          break;
 
diff --git a/src/mesa/drivers/dri/i965/brw_shader.cpp b/src/mesa/drivers/dri/i965/brw_shader.cpp
index 9550a62..e64adb1 100644
--- a/src/mesa/drivers/dri/i965/brw_shader.cpp
+++ b/src/mesa/drivers/dri/i965/brw_shader.cpp
@@ -429,6 +429,8 @@ brw_instruction_name(enum opcode op)
       return "gen8_urb_write_simd8_masked_per_slot";
    case SHADER_OPCODE_URB_READ_SIMD8:
       return "urb_read_simd8";
+   case SHADER_OPCODE_URB_READ_SIMD8_PER_SLOT:
+      return "urb_read_simd8_per_slot";
 
    case SHADER_OPCODE_FIND_LIVE_CHANNEL:
       return "find_live_channel";
-- 
2.6.2