On Sun, Nov 8, 2015 at 9:12 PM, Rob Clark <robdcl...@gmail.com> wrote: > --- > src/gallium/include/pipe/p_defines.h | 1 + > src/gallium/include/pipe/p_state.h | 7 +++++++ > 2 files changed, 8 insertions(+) > > diff --git a/src/gallium/include/pipe/p_defines.h > b/src/gallium/include/pipe/p_defines.h > index 0a9d98d..572461f 100644 > --- a/src/gallium/include/pipe/p_defines.h > +++ b/src/gallium/include/pipe/p_defines.h > @@ -718,6 +718,7 @@ enum pipe_shader_ir > PIPE_SHADER_IR_TGSI = 0, > PIPE_SHADER_IR_LLVM, > PIPE_SHADER_IR_NATIVE, > + PIPE_SHADER_IR_NIR, > }; > > /** > diff --git a/src/gallium/include/pipe/p_state.h > b/src/gallium/include/pipe/p_state.h > index f1c4b49..7eee709 100644 > --- a/src/gallium/include/pipe/p_state.h > +++ b/src/gallium/include/pipe/p_state.h > @@ -221,6 +221,12 @@ struct pipe_stream_output_info > * > * TODO pipe_compute_state should probably get similar treatment to handle > * multiple IR's in a cleaner way.. > + * > + * NOTE: since the nir_shader is reference counted, the semantics are a bit > + * different from create_xyz_state(ir=TGSI). The driver takes ownership of > + * the nir_shader (and must nir_shader_unref()) at some point. If state > + * trackers need to hang on to the IR (for example, variant management), it > + * should increment the refcnt before calling create_xyz_shader(ir=NIR).
Sounds good. Marek _______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
