i965/fs: Drop the emit(fs_inst) overload.

Using this emit function implicitly creates three copies, which
is pointlessly inefficient.

1. Code creates the original instruction.
2. Calling emit(fs_inst) copies it into the function.
3. It then allocates a new fs_inst and copies it into that.

The second could be eliminated by changing the signature to

   fs_inst(const fs_inst &)

but that wouldn't eliminate the third.  Making callers heap allocate the
instruction and call emit(fs_inst *) allows us to just use the original
one, with no extra copies, and isn't much more of a burden.

Reviewed-by: Matt Turner <mattst88@gmail.com>
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>

diff --git a/src/mesa/drivers/dri/i965/brw_fs.cpp b/src/mesa/drivers/dri/i965/brw_fs.cpp
index 65ee2f021e4a3b3ee05ad268d5ccf37b88a61800..a88c621d047fde7283ae49f802f467cd3f7ffe82 100644 (file)
--- a/src/mesa/drivers/dri/i965/brw_fs.cpp
+++ b/src/mesa/drivers/dri/i965/brw_fs.cpp
@@ -642,8 +642,8 @@ fs_visitor::emit_shader_time_write(enum shader_time_shader_type type,
    else
       payload = fs_reg(this, glsl_type::uint_type);
 
-   emit(fs_inst(SHADER_OPCODE_SHADER_TIME_ADD,
-                fs_reg(), payload, offset, value));
+   emit(new(mem_ctx) fs_inst(SHADER_OPCODE_SHADER_TIME_ADD,
+                             fs_reg(), payload, offset, value));
 }
 
 void
@@ -672,32 +672,32 @@ fs_visitor::fail(const char *format, ...)
 fs_inst *
 fs_visitor::emit(enum opcode opcode)
 {
-   return emit(fs_inst(opcode));
+   return emit(new(mem_ctx) fs_inst(opcode));
 }
 
 fs_inst *
 fs_visitor::emit(enum opcode opcode, fs_reg dst)
 {
-   return emit(fs_inst(opcode, dst));
+   return emit(new(mem_ctx) fs_inst(opcode, dst));
 }
 
 fs_inst *
 fs_visitor::emit(enum opcode opcode, fs_reg dst, fs_reg src0)
 {
-   return emit(fs_inst(opcode, dst, src0));
+   return emit(new(mem_ctx) fs_inst(opcode, dst, src0));
 }
 
 fs_inst *
 fs_visitor::emit(enum opcode opcode, fs_reg dst, fs_reg src0, fs_reg src1)
 {
-   return emit(fs_inst(opcode, dst, src0, src1));
+   return emit(new(mem_ctx) fs_inst(opcode, dst, src0, src1));
 }
 
 fs_inst *
 fs_visitor::emit(enum opcode opcode, fs_reg dst,
                  fs_reg src0, fs_reg src1, fs_reg src2)
 {
-   return emit(fs_inst(opcode, dst, src0, src1, src2));
+   return emit(new(mem_ctx) fs_inst(opcode, dst, src0, src1, src2));
 }
 
 void

diff --git a/src/mesa/drivers/dri/i965/brw_fs.h b/src/mesa/drivers/dri/i965/brw_fs.h
index 1d95ed722016c15fc3f5c45584425cbb8da8e241..fff7d8e474d90e5d20f6193a009a844f31827fd8 100644 (file)
--- a/src/mesa/drivers/dri/i965/brw_fs.h
+++ b/src/mesa/drivers/dri/i965/brw_fs.h
@@ -284,7 +284,6 @@ public:
 
    bool can_do_source_mods(fs_inst *inst);
 
-   fs_inst *emit(fs_inst inst);
    fs_inst *emit(fs_inst *inst);
    void emit(exec_list list);
 

diff --git a/src/mesa/drivers/dri/i965/brw_fs_visitor.cpp b/src/mesa/drivers/dri/i965/brw_fs_visitor.cpp
index 9a0090dcc384d93972068a6d43d00628a9901e02..9b090be7bde62831187f8fb98af892a0e866dd41 100644 (file)
--- a/src/mesa/drivers/dri/i965/brw_fs_visitor.cpp
+++ b/src/mesa/drivers/dri/i965/brw_fs_visitor.cpp
@@ -743,8 +743,8 @@ fs_visitor::visit(ir_expression *ir)
          packed_consts.type = result.type;
 
          fs_reg const_offset_reg = fs_reg(const_offset->value.u[0] & ~15);
-         emit(fs_inst(FS_OPCODE_UNIFORM_PULL_CONSTANT_LOAD,
-                      packed_consts, surf_index, const_offset_reg));
+         emit(new(mem_ctx) fs_inst(FS_OPCODE_UNIFORM_PULL_CONSTANT_LOAD,
+                                   packed_consts, surf_index, const_offset_reg));
 
          for (int i = 0; i < ir->type->vector_elements; i++) {
             packed_consts.set_smear(const_offset->value.u[0] % 16 / 4 + i);
@@ -2399,9 +2399,10 @@ fs_visitor::emit_untyped_atomic(unsigned atomic_op, unsigned surf_index,
    }
 
    /* Emit the instruction. */
-   fs_inst inst(SHADER_OPCODE_UNTYPED_ATOMIC, dst, atomic_op, surf_index);
-   inst.base_mrf = 0;
-   inst.mlen = mlen;
+   fs_inst *inst = new(mem_ctx) fs_inst(SHADER_OPCODE_UNTYPED_ATOMIC, dst,
+                                        atomic_op, surf_index);
+   inst->base_mrf = 0;
+   inst->mlen = mlen;
    emit(inst);
 }
 
@@ -2432,21 +2433,13 @@ fs_visitor::emit_untyped_surface_read(unsigned surf_index, fs_reg dst,
    mlen += operand_len;
 
    /* Emit the instruction. */
-   fs_inst inst(SHADER_OPCODE_UNTYPED_SURFACE_READ, dst, surf_index);
-   inst.base_mrf = 0;
-   inst.mlen = mlen;
+   fs_inst *inst = new(mem_ctx)
+      fs_inst(SHADER_OPCODE_UNTYPED_SURFACE_READ, dst, surf_index);
+   inst->base_mrf = 0;
+   inst->mlen = mlen;
    emit(inst);
 }
 
-fs_inst *
-fs_visitor::emit(fs_inst inst)
-{
-   fs_inst *list_inst = new(mem_ctx) fs_inst;
-   *list_inst = inst;
-   emit(list_inst);
-   return list_inst;
-}
-
 fs_inst *
 fs_visitor::emit(fs_inst *inst)
 {