Cycles: change __device and similar qualifiers to ccl_device in kernel code.

This to avoids build conflicts with libc++ on FreeBSD, these __ prefixed values
are reserved for compilers. I apologize to anyone who has patches or branches
and has to go through the pain of merging this change, it may be easiest to do
these same replacements in your code and then apply/merge the patch.

Ref T37477.

18 files changed, 126 insertions, 126 deletions

diff --git a/intern/cycles/kernel/closure/bsdf.h b/intern/cycles/kernel/closure/bsdf.h
index 86fea48760f..b3141d1154f 100644
--- a/intern/cycles/kernel/closure/bsdf.h
+++ b/intern/cycles/kernel/closure/bsdf.h
@@ -35,7 +35,7 @@
 
 CCL_NAMESPACE_BEGIN
 
-__device int bsdf_sample(KernelGlobals *kg, const ShaderData *sd, const ShaderClosure *sc, float randu, float randv, float3 *eval, float3 *omega_in, differential3 *domega_in, float *pdf)
+ccl_device int bsdf_sample(KernelGlobals *kg, const ShaderData *sd, const ShaderClosure *sc, float randu, float randv, float3 *eval, float3 *omega_in, differential3 *domega_in, float *pdf)
 {
 	int label;
 
@@ -132,7 +132,7 @@ __device int bsdf_sample(KernelGlobals *kg, const ShaderData *sd, const ShaderCl
 	return label;
 }
 
-__device float3 bsdf_eval(KernelGlobals *kg, const ShaderData *sd, const ShaderClosure *sc, const float3 omega_in, float *pdf)
+ccl_device float3 bsdf_eval(KernelGlobals *kg, const ShaderData *sd, const ShaderClosure *sc, const float3 omega_in, float *pdf)
 {
 	float3 eval;
 
@@ -275,7 +275,7 @@ __device float3 bsdf_eval(KernelGlobals *kg, const ShaderData *sd, const ShaderC
 	return eval;
 }
 
-__device void bsdf_blur(KernelGlobals *kg, ShaderClosure *sc, float roughness)
+ccl_device void bsdf_blur(KernelGlobals *kg, ShaderClosure *sc, float roughness)
 {
 #ifdef __OSL__
 	if(kg->osl && sc->prim) {
diff --git a/intern/cycles/kernel/closure/bsdf_ashikhmin_velvet.h b/intern/cycles/kernel/closure/bsdf_ashikhmin_velvet.h
index 94bc6eb0dc5..3631f90bf8c 100644
--- a/intern/cycles/kernel/closure/bsdf_ashikhmin_velvet.h
+++ b/intern/cycles/kernel/closure/bsdf_ashikhmin_velvet.h
@@ -35,7 +35,7 @@
 
 CCL_NAMESPACE_BEGIN
 
-__device int bsdf_ashikhmin_velvet_setup(ShaderClosure *sc)
+ccl_device int bsdf_ashikhmin_velvet_setup(ShaderClosure *sc)
 {
 	float sigma = fmaxf(sc->data0, 0.01f);
 	sc->data0 = 1.0f/(sigma * sigma); /* m_invsigma2 */
@@ -45,11 +45,11 @@ __device int bsdf_ashikhmin_velvet_setup(ShaderClosure *sc)
 	return SD_BSDF|SD_BSDF_HAS_EVAL;
 }
 
-__device void bsdf_ashikhmin_velvet_blur(ShaderClosure *sc, float roughness)
+ccl_device void bsdf_ashikhmin_velvet_blur(ShaderClosure *sc, float roughness)
 {
 }
 
-__device float3 bsdf_ashikhmin_velvet_eval_reflect(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
+ccl_device float3 bsdf_ashikhmin_velvet_eval_reflect(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
 {
 	float m_invsigma2 = sc->data0;
 	float3 N = sc->N;
@@ -87,12 +87,12 @@ __device float3 bsdf_ashikhmin_velvet_eval_reflect(const ShaderClosure *sc, cons
 	return make_float3(0, 0, 0);
 }
 
-__device float3 bsdf_ashikhmin_velvet_eval_transmit(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
+ccl_device float3 bsdf_ashikhmin_velvet_eval_transmit(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
 {
 	return make_float3(0.0f, 0.0f, 0.0f);
 }
 
-__device int bsdf_ashikhmin_velvet_sample(const ShaderClosure *sc, float3 Ng, float3 I, float3 dIdx, float3 dIdy, float randu, float randv, float3 *eval, float3 *omega_in, float3 *domega_in_dx, float3 *domega_in_dy, float *pdf)
+ccl_device int bsdf_ashikhmin_velvet_sample(const ShaderClosure *sc, float3 Ng, float3 I, float3 dIdx, float3 dIdy, float randu, float randv, float3 *eval, float3 *omega_in, float3 *domega_in_dx, float3 *domega_in_dy, float *pdf)
 {
 	float m_invsigma2 = sc->data0;
 	float3 N = sc->N;
diff --git a/intern/cycles/kernel/closure/bsdf_diffuse.h b/intern/cycles/kernel/closure/bsdf_diffuse.h
index 46318ecd138..949fe869549 100644
--- a/intern/cycles/kernel/closure/bsdf_diffuse.h
+++ b/intern/cycles/kernel/closure/bsdf_diffuse.h
@@ -37,17 +37,17 @@ CCL_NAMESPACE_BEGIN
 
 /* DIFFUSE */
 
-__device int bsdf_diffuse_setup(ShaderClosure *sc)
+ccl_device int bsdf_diffuse_setup(ShaderClosure *sc)
 {
 	sc->type = CLOSURE_BSDF_DIFFUSE_ID;
 	return SD_BSDF|SD_BSDF_HAS_EVAL;
 }
 
-__device void bsdf_diffuse_blur(ShaderClosure *sc, float roughness)
+ccl_device void bsdf_diffuse_blur(ShaderClosure *sc, float roughness)
 {
 }
 
-__device float3 bsdf_diffuse_eval_reflect(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
+ccl_device float3 bsdf_diffuse_eval_reflect(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
 {
 	float3 N = sc->N;
 
@@ -56,12 +56,12 @@ __device float3 bsdf_diffuse_eval_reflect(const ShaderClosure *sc, const float3
 	return make_float3(cos_pi, cos_pi, cos_pi);
 }
 
-__device float3 bsdf_diffuse_eval_transmit(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
+ccl_device float3 bsdf_diffuse_eval_transmit(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
 {
 	return make_float3(0.0f, 0.0f, 0.0f);
 }
 
-__device int bsdf_diffuse_sample(const ShaderClosure *sc, float3 Ng, float3 I, float3 dIdx, float3 dIdy, float randu, float randv, float3 *eval, float3 *omega_in, float3 *domega_in_dx, float3 *domega_in_dy, float *pdf)
+ccl_device int bsdf_diffuse_sample(const ShaderClosure *sc, float3 Ng, float3 I, float3 dIdx, float3 dIdy, float randu, float randv, float3 *eval, float3 *omega_in, float3 *domega_in_dx, float3 *domega_in_dy, float *pdf)
 {
 	float3 N = sc->N;
 
@@ -84,22 +84,22 @@ __device int bsdf_diffuse_sample(const ShaderClosure *sc, float3 Ng, float3 I, f
 
 /* TRANSLUCENT */
 
-__device int bsdf_translucent_setup(ShaderClosure *sc)
+ccl_device int bsdf_translucent_setup(ShaderClosure *sc)
 {
 	sc->type = CLOSURE_BSDF_TRANSLUCENT_ID;
 	return SD_BSDF|SD_BSDF_HAS_EVAL;
 }
 
-__device void bsdf_translucent_blur(ShaderClosure *sc, float roughness)
+ccl_device void bsdf_translucent_blur(ShaderClosure *sc, float roughness)
 {
 }
 
-__device float3 bsdf_translucent_eval_reflect(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
+ccl_device float3 bsdf_translucent_eval_reflect(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
 {
 	return make_float3(0.0f, 0.0f, 0.0f);
 }
 
-__device float3 bsdf_translucent_eval_transmit(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
+ccl_device float3 bsdf_translucent_eval_transmit(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
 {
 	float3 N = sc->N;
 
@@ -108,12 +108,12 @@ __device float3 bsdf_translucent_eval_transmit(const ShaderClosure *sc, const fl
 	return make_float3 (cos_pi, cos_pi, cos_pi);
 }
 
-__device float bsdf_translucent_albedo(const ShaderClosure *sc, const float3 I)
+ccl_device float bsdf_translucent_albedo(const ShaderClosure *sc, const float3 I)
 {
 	return 1.0f;
 }
 
-__device int bsdf_translucent_sample(const ShaderClosure *sc, float3 Ng, float3 I, float3 dIdx, float3 dIdy, float randu, float randv, float3 *eval, float3 *omega_in, float3 *domega_in_dx, float3 *domega_in_dy, float *pdf)
+ccl_device int bsdf_translucent_sample(const ShaderClosure *sc, float3 Ng, float3 I, float3 dIdx, float3 dIdy, float randu, float randv, float3 *eval, float3 *omega_in, float3 *domega_in_dx, float3 *domega_in_dy, float *pdf)
 {
 	float3 N = sc->N;
 
diff --git a/intern/cycles/kernel/closure/bsdf_diffuse_ramp.h b/intern/cycles/kernel/closure/bsdf_diffuse_ramp.h
index 2e43e16693f..b856774375f 100644
--- a/intern/cycles/kernel/closure/bsdf_diffuse_ramp.h
+++ b/intern/cycles/kernel/closure/bsdf_diffuse_ramp.h
@@ -35,7 +35,7 @@
 
 CCL_NAMESPACE_BEGIN
 
-__device float3 bsdf_diffuse_ramp_get_color(const ShaderClosure *sc, const float3 colors[8], float pos)
+ccl_device float3 bsdf_diffuse_ramp_get_color(const ShaderClosure *sc, const float3 colors[8], float pos)
 {
 	int MAXCOLORS = 8;
 	
@@ -49,17 +49,17 @@ __device float3 bsdf_diffuse_ramp_get_color(const ShaderClosure *sc, const float
 	return colors[ipos] * (1.0f - offset) + colors[ipos+1] * offset;
 }
 
-__device int bsdf_diffuse_ramp_setup(ShaderClosure *sc)
+ccl_device int bsdf_diffuse_ramp_setup(ShaderClosure *sc)
 {
 	sc->type = CLOSURE_BSDF_DIFFUSE_RAMP_ID;
 	return SD_BSDF | SD_BSDF_HAS_EVAL;
 }
 
-__device void bsdf_diffuse_ramp_blur(ShaderClosure *sc, float roughness)
+ccl_device void bsdf_diffuse_ramp_blur(ShaderClosure *sc, float roughness)
 {
 }
 
-__device float3 bsdf_diffuse_ramp_eval_reflect(const ShaderClosure *sc, const float3 colors[8], const float3 I, const float3 omega_in, float *pdf)
+ccl_device float3 bsdf_diffuse_ramp_eval_reflect(const ShaderClosure *sc, const float3 colors[8], const float3 I, const float3 omega_in, float *pdf)
 {
 	float3 N = sc->N;
 
@@ -68,12 +68,12 @@ __device float3 bsdf_diffuse_ramp_eval_reflect(const ShaderClosure *sc, const fl
 	return bsdf_diffuse_ramp_get_color(sc, colors, cos_pi) * M_1_PI_F;
 }
 
-__device float3 bsdf_diffuse_ramp_eval_transmit(const ShaderClosure *sc, const float3 colors[8], const float3 I, const float3 omega_in, float *pdf)
+ccl_device float3 bsdf_diffuse_ramp_eval_transmit(const ShaderClosure *sc, const float3 colors[8], const float3 I, const float3 omega_in, float *pdf)
 {
 	return make_float3(0.0f, 0.0f, 0.0f);
 }
 
-__device int bsdf_diffuse_ramp_sample(const ShaderClosure *sc, const float3 colors[8], float3 Ng, float3 I, float3 dIdx, float3 dIdy, float randu, float randv, float3 *eval, float3 *omega_in, float3 *domega_in_dx, float3 *domega_in_dy, float *pdf)
+ccl_device int bsdf_diffuse_ramp_sample(const ShaderClosure *sc, const float3 colors[8], float3 Ng, float3 I, float3 dIdx, float3 dIdy, float randu, float randv, float3 *eval, float3 *omega_in, float3 *domega_in_dx, float3 *domega_in_dy, float *pdf)
 {
 	float3 N = sc->N;
 
diff --git a/intern/cycles/kernel/closure/bsdf_hair.h b/intern/cycles/kernel/closure/bsdf_hair.h
index 5825d2637ba..163e7cc5ee2 100644
--- a/intern/cycles/kernel/closure/bsdf_hair.h
+++ b/intern/cycles/kernel/closure/bsdf_hair.h
@@ -36,15 +36,15 @@
 CCL_NAMESPACE_BEGIN
 
 
-__device void bsdf_hair_reflection_blur(ShaderClosure *sc, float roughness)
+ccl_device void bsdf_hair_reflection_blur(ShaderClosure *sc, float roughness)
 {
 }
 
-__device void bsdf_hair_transmission_blur(ShaderClosure *sc, float roughness)
+ccl_device void bsdf_hair_transmission_blur(ShaderClosure *sc, float roughness)
 {
 }
 
-__device int bsdf_hair_reflection_setup(ShaderClosure *sc)
+ccl_device int bsdf_hair_reflection_setup(ShaderClosure *sc)
 {
 	sc->type = CLOSURE_BSDF_HAIR_REFLECTION_ID;
 	sc->data0 = clamp(sc->data0, 0.001f, 1.0f);
@@ -52,7 +52,7 @@ __device int bsdf_hair_reflection_setup(ShaderClosure *sc)
 	return SD_BSDF|SD_BSDF_HAS_EVAL|SD_BSDF_GLOSSY;
 }
 
-__device int bsdf_hair_transmission_setup(ShaderClosure *sc)
+ccl_device int bsdf_hair_transmission_setup(ShaderClosure *sc)
 {
 	sc->type = CLOSURE_BSDF_HAIR_TRANSMISSION_ID;
 	sc->data0 = clamp(sc->data0, 0.001f, 1.0f);
@@ -60,7 +60,7 @@ __device int bsdf_hair_transmission_setup(ShaderClosure *sc)
 	return SD_BSDF|SD_BSDF_HAS_EVAL|SD_BSDF_GLOSSY;
 }
 
-__device float3 bsdf_hair_reflection_eval_reflect(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
+ccl_device float3 bsdf_hair_reflection_eval_reflect(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
 {
 #ifdef __HAIR__
 	float offset = sc->offset;
@@ -106,18 +106,18 @@ __device float3 bsdf_hair_reflection_eval_reflect(const ShaderClosure *sc, const
 	return make_float3(*pdf, *pdf, *pdf);
 }
 
-__device float3 bsdf_hair_transmission_eval_reflect(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
+ccl_device float3 bsdf_hair_transmission_eval_reflect(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
 {
 	return make_float3(0.0f, 0.0f, 0.0f);
 }
 
 
-__device float3 bsdf_hair_reflection_eval_transmit(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
+ccl_device float3 bsdf_hair_reflection_eval_transmit(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
 {
 	return make_float3(0.0f, 0.0f, 0.0f);
 }
 
-__device float3 bsdf_hair_transmission_eval_transmit(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
+ccl_device float3 bsdf_hair_transmission_eval_transmit(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
 {
 #ifdef __HAIR__
 	float offset = sc->offset;
@@ -163,7 +163,7 @@ __device float3 bsdf_hair_transmission_eval_transmit(const ShaderClosure *sc, co
 	return make_float3(*pdf, *pdf, *pdf);
 }
 
-__device int bsdf_hair_reflection_sample(const ShaderClosure *sc, float3 Ng, float3 I, float3 dIdx, float3 dIdy, float randu, float randv, float3 *eval, float3 *omega_in, float3 *domega_in_dx, float3 *domega_in_dy, float *pdf)
+ccl_device int bsdf_hair_reflection_sample(const ShaderClosure *sc, float3 Ng, float3 I, float3 dIdx, float3 dIdy, float randu, float randv, float3 *eval, float3 *omega_in, float3 *domega_in_dx, float3 *domega_in_dy, float *pdf)
 {
 #ifdef __HAIR__
 	float offset = sc->offset;
@@ -218,7 +218,7 @@ __device int bsdf_hair_reflection_sample(const ShaderClosure *sc, float3 Ng, flo
 	return LABEL_REFLECT|LABEL_GLOSSY;
 }
 
-__device int bsdf_hair_transmission_sample(const ShaderClosure *sc, float3 Ng, float3 I, float3 dIdx, float3 dIdy, float randu, float randv, float3 *eval, float3 *omega_in, float3 *domega_in_dx, float3 *domega_in_dy, float *pdf)
+ccl_device int bsdf_hair_transmission_sample(const ShaderClosure *sc, float3 Ng, float3 I, float3 dIdx, float3 dIdy, float randu, float randv, float3 *eval, float3 *omega_in, float3 *domega_in_dx, float3 *domega_in_dy, float *pdf)
 {
 #ifdef __HAIR__
 	float offset = sc->offset;
diff --git a/intern/cycles/kernel/closure/bsdf_microfacet.h b/intern/cycles/kernel/closure/bsdf_microfacet.h
index b159f585831..737cffb0f18 100644
--- a/intern/cycles/kernel/closure/bsdf_microfacet.h
+++ b/intern/cycles/kernel/closure/bsdf_microfacet.h
@@ -37,7 +37,7 @@ CCL_NAMESPACE_BEGIN
 
 /* GGX */
 
-__device int bsdf_microfacet_ggx_setup(ShaderClosure *sc)
+ccl_device int bsdf_microfacet_ggx_setup(ShaderClosure *sc)
 {
 	sc->data0 = clamp(sc->data0, 0.0f, 1.0f); /* m_ag */
 	
@@ -46,7 +46,7 @@ __device int bsdf_microfacet_ggx_setup(ShaderClosure *sc)
 	return SD_BSDF|SD_BSDF_HAS_EVAL|SD_BSDF_GLOSSY;
 }
 
-__device int bsdf_microfacet_ggx_refraction_setup(ShaderClosure *sc)
+ccl_device int bsdf_microfacet_ggx_refraction_setup(ShaderClosure *sc)
 {
 	sc->data0 = clamp(sc->data0, 0.0f, 1.0f); /* m_ag */
 
@@ -55,12 +55,12 @@ __device int bsdf_microfacet_ggx_refraction_setup(ShaderClosure *sc)
 	return SD_BSDF|SD_BSDF_HAS_EVAL|SD_BSDF_GLOSSY;
 }
 
-__device void bsdf_microfacet_ggx_blur(ShaderClosure *sc, float roughness)
+ccl_device void bsdf_microfacet_ggx_blur(ShaderClosure *sc, float roughness)
 {
 	sc->data0 = fmaxf(roughness, sc->data0); /* m_ag */
 }
 
-__device float3 bsdf_microfacet_ggx_eval_reflect(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
+ccl_device float3 bsdf_microfacet_ggx_eval_reflect(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
 {
 	float m_ag = max(sc->data0, 1e-4f);
 	int m_refractive = sc->type == CLOSURE_BSDF_MICROFACET_GGX_REFRACTION_ID;
@@ -97,7 +97,7 @@ __device float3 bsdf_microfacet_ggx_eval_reflect(const ShaderClosure *sc, const
 	return make_float3 (0, 0, 0);
 }
 
-__device float3 bsdf_microfacet_ggx_eval_transmit(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
+ccl_device float3 bsdf_microfacet_ggx_eval_transmit(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
 {
 	float m_ag = max(sc->data0, 1e-4f);
 	float m_eta = sc->data1;
@@ -134,7 +134,7 @@ __device float3 bsdf_microfacet_ggx_eval_transmit(const ShaderClosure *sc, const
 	return make_float3 (out, out, out);
 }
 
-__device int bsdf_microfacet_ggx_sample(const ShaderClosure *sc, float3 Ng, float3 I, float3 dIdx, float3 dIdy, float randu, float randv, float3 *eval, float3 *omega_in, float3 *domega_in_dx, float3 *domega_in_dy, float *pdf)
+ccl_device int bsdf_microfacet_ggx_sample(const ShaderClosure *sc, float3 Ng, float3 I, float3 dIdx, float3 dIdy, float randu, float randv, float3 *eval, float3 *omega_in, float3 *domega_in_dx, float3 *domega_in_dy, float *pdf)
 {
 	float m_ag = sc->data0;
 	int m_refractive = sc->type == CLOSURE_BSDF_MICROFACET_GGX_REFRACTION_ID;
@@ -255,7 +255,7 @@ __device int bsdf_microfacet_ggx_sample(const ShaderClosure *sc, float3 Ng, floa
 
 /* BECKMANN */
 
-__device int bsdf_microfacet_beckmann_setup(ShaderClosure *sc)
+ccl_device int bsdf_microfacet_beckmann_setup(ShaderClosure *sc)
 {
 	sc->data0 = clamp(sc->data0, 0.0f, 1.0f); /* m_ab */
 
@@ -263,7 +263,7 @@ __device int bsdf_microfacet_beckmann_setup(ShaderClosure *sc)
 	return SD_BSDF|SD_BSDF_HAS_EVAL|SD_BSDF_GLOSSY;
 }
 
-__device int bsdf_microfacet_beckmann_refraction_setup(ShaderClosure *sc)
+ccl_device int bsdf_microfacet_beckmann_refraction_setup(ShaderClosure *sc)
 {
 	sc->data0 = clamp(sc->data0, 0.0f, 1.0f); /* m_ab */
 
@@ -271,12 +271,12 @@ __device int bsdf_microfacet_beckmann_refraction_setup(ShaderClosure *sc)
 	return SD_BSDF|SD_BSDF_HAS_EVAL|SD_BSDF_GLOSSY;
 }
 
-__device void bsdf_microfacet_beckmann_blur(ShaderClosure *sc, float roughness)
+ccl_device void bsdf_microfacet_beckmann_blur(ShaderClosure *sc, float roughness)
 {
 	sc->data0 = fmaxf(roughness, sc->data0); /* m_ab */
 }
 
-__device float3 bsdf_microfacet_beckmann_eval_reflect(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
+ccl_device float3 bsdf_microfacet_beckmann_eval_reflect(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
 {
 	float m_ab = max(sc->data0, 1e-4f);
 	int m_refractive = sc->type == CLOSURE_BSDF_MICROFACET_BECKMANN_REFRACTION_ID;
@@ -315,7 +315,7 @@ __device float3 bsdf_microfacet_beckmann_eval_reflect(const ShaderClosure *sc, c
 	return make_float3 (0, 0, 0);
 }
 
-__device float3 bsdf_microfacet_beckmann_eval_transmit(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
+ccl_device float3 bsdf_microfacet_beckmann_eval_transmit(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
 {
 	float m_ab = max(sc->data0, 1e-4f);
 	float m_eta = sc->data1;
@@ -354,7 +354,7 @@ __device float3 bsdf_microfacet_beckmann_eval_transmit(const ShaderClosure *sc,
 	return make_float3 (out, out, out);
 }
 
-__device int bsdf_microfacet_beckmann_sample(const ShaderClosure *sc, float3 Ng, float3 I, float3 dIdx, float3 dIdy, float randu, float randv, float3 *eval, float3 *omega_in, float3 *domega_in_dx, float3 *domega_in_dy, float *pdf)
+ccl_device int bsdf_microfacet_beckmann_sample(const ShaderClosure *sc, float3 Ng, float3 I, float3 dIdx, float3 dIdy, float randu, float randv, float3 *eval, float3 *omega_in, float3 *domega_in_dx, float3 *domega_in_dy, float *pdf)
 {
 	float m_ab = sc->data0;
 	int m_refractive = sc->type == CLOSURE_BSDF_MICROFACET_BECKMANN_REFRACTION_ID;
diff --git a/intern/cycles/kernel/closure/bsdf_oren_nayar.h b/intern/cycles/kernel/closure/bsdf_oren_nayar.h
index c6c6811c007..6f685d5eeea 100644
--- a/intern/cycles/kernel/closure/bsdf_oren_nayar.h
+++ b/intern/cycles/kernel/closure/bsdf_oren_nayar.h
@@ -19,7 +19,7 @@
 
 CCL_NAMESPACE_BEGIN
 
-__device float3 bsdf_oren_nayar_get_intensity(const ShaderClosure *sc, float3 n, float3 v, float3 l)
+ccl_device float3 bsdf_oren_nayar_get_intensity(const ShaderClosure *sc, float3 n, float3 v, float3 l)
 {
 	float nl = max(dot(n, l), 0.0f);
 	float nv = max(dot(n, v), 0.0f);
@@ -31,7 +31,7 @@ __device float3 bsdf_oren_nayar_get_intensity(const ShaderClosure *sc, float3 n,
 	return make_float3(is, is, is);
 }
 
-__device int bsdf_oren_nayar_setup(ShaderClosure *sc)
+ccl_device int bsdf_oren_nayar_setup(ShaderClosure *sc)
 {
 	float sigma = sc->data0;
 
@@ -47,11 +47,11 @@ __device int bsdf_oren_nayar_setup(ShaderClosure *sc)
 	return SD_BSDF | SD_BSDF_HAS_EVAL;
 }
 
-__device void bsdf_oren_nayar_blur(ShaderClosure *sc, float roughness)
+ccl_device void bsdf_oren_nayar_blur(ShaderClosure *sc, float roughness)
 {
 }
 
-__device float3 bsdf_oren_nayar_eval_reflect(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
+ccl_device float3 bsdf_oren_nayar_eval_reflect(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
 {
 	if (dot(sc->N, omega_in) > 0.0f) {
 		*pdf = 0.5f * M_1_PI_F;
@@ -63,12 +63,12 @@ __device float3 bsdf_oren_nayar_eval_reflect(const ShaderClosure *sc, const floa
 	}
 }
 
-__device float3 bsdf_oren_nayar_eval_transmit(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
+ccl_device float3 bsdf_oren_nayar_eval_transmit(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
 {
 	return make_float3(0.0f, 0.0f, 0.0f);
 }
 
-__device int bsdf_oren_nayar_sample(const ShaderClosure *sc, float3 Ng, float3 I, float3 dIdx, float3 dIdy, float randu, float randv, float3 *eval, float3 *omega_in, float3 *domega_in_dx, float3 *domega_in_dy, float *pdf)
+ccl_device int bsdf_oren_nayar_sample(const ShaderClosure *sc, float3 Ng, float3 I, float3 dIdx, float3 dIdy, float randu, float randv, float3 *eval, float3 *omega_in, float3 *domega_in_dx, float3 *domega_in_dy, float *pdf)
 {
 	sample_uniform_hemisphere(sc->N, randu, randv, omega_in, pdf);
 
diff --git a/intern/cycles/kernel/closure/bsdf_phong_ramp.h b/intern/cycles/kernel/closure/bsdf_phong_ramp.h
index 1e332933287..219c5aea159 100644
--- a/intern/cycles/kernel/closure/bsdf_phong_ramp.h
+++ b/intern/cycles/kernel/closure/bsdf_phong_ramp.h
@@ -35,7 +35,7 @@
 
 CCL_NAMESPACE_BEGIN
 
-__device float3 bsdf_phong_ramp_get_color(const ShaderClosure *sc, const float3 colors[8], float pos)
+ccl_device float3 bsdf_phong_ramp_get_color(const ShaderClosure *sc, const float3 colors[8], float pos)
 {
 	int MAXCOLORS = 8;
 	
@@ -49,7 +49,7 @@ __device float3 bsdf_phong_ramp_get_color(const ShaderClosure *sc, const float3
 	return colors[ipos] * (1.0f - offset) + colors[ipos+1] * offset;
 }
 
-__device int bsdf_phong_ramp_setup(ShaderClosure *sc)
+ccl_device int bsdf_phong_ramp_setup(ShaderClosure *sc)
 {
 	sc->data0 = max(sc->data0, 0.0f);
 	
@@ -57,11 +57,11 @@ __device int bsdf_phong_ramp_setup(ShaderClosure *sc)
 	return SD_BSDF | SD_BSDF_HAS_EVAL | SD_BSDF_GLOSSY;
 }
 
-__device void bsdf_phong_ramp_blur(ShaderClosure *sc, float roughness)
+ccl_device void bsdf_phong_ramp_blur(ShaderClosure *sc, float roughness)
 {
 }
 
-__device float3 bsdf_phong_ramp_eval_reflect(const ShaderClosure *sc, const float3 colors[8], const float3 I, const float3 omega_in, float *pdf)
+ccl_device float3 bsdf_phong_ramp_eval_reflect(const ShaderClosure *sc, const float3 colors[8], const float3 I, const float3 omega_in, float *pdf)
 {
 	float m_exponent = sc->data0;
 	float cosNI = dot(sc->N, omega_in);
@@ -83,12 +83,12 @@ __device float3 bsdf_phong_ramp_eval_reflect(const ShaderClosure *sc, const floa
 	return make_float3(0.0f, 0.0f, 0.0f);
 }
 
-__device float3 bsdf_phong_ramp_eval_transmit(const ShaderClosure *sc, const float3 colors[8], const float3 I, const float3 omega_in, float *pdf)
+ccl_device float3 bsdf_phong_ramp_eval_transmit(const ShaderClosure *sc, const float3 colors[8], const float3 I, const float3 omega_in, float *pdf)
 {
 	return make_float3(0.0f, 0.0f, 0.0f);
 }
 
-__device int bsdf_phong_ramp_sample(const ShaderClosure *sc, const float3 colors[8], float3 Ng, float3 I, float3 dIdx, float3 dIdy, float randu, float randv, float3 *eval, float3 *omega_in, float3 *domega_in_dx, float3 *domega_in_dy, float *pdf)
+ccl_device int bsdf_phong_ramp_sample(const ShaderClosure *sc, const float3 colors[8], float3 Ng, float3 I, float3 dIdx, float3 dIdy, float randu, float randv, float3 *eval, float3 *omega_in, float3 *domega_in_dx, float3 *domega_in_dy, float *pdf)
 {
 	float cosNO = dot(sc->N, I);
 	float m_exponent = sc->data0;
diff --git a/intern/cycles/kernel/closure/bsdf_reflection.h b/intern/cycles/kernel/closure/bsdf_reflection.h
index 7715aac936f..0baccdf155c 100644
--- a/intern/cycles/kernel/closure/bsdf_reflection.h
+++ b/intern/cycles/kernel/closure/bsdf_reflection.h
@@ -37,27 +37,27 @@ CCL_NAMESPACE_BEGIN
 
 /* REFLECTION */
 
-__device int bsdf_reflection_setup(ShaderClosure *sc)
+ccl_device int bsdf_reflection_setup(ShaderClosure *sc)
 {
 	sc->type = CLOSURE_BSDF_REFLECTION_ID;
 	return SD_BSDF;
 }
 
-__device void bsdf_reflection_blur(ShaderClosure *sc, float roughness)
+ccl_device void bsdf_reflection_blur(ShaderClosure *sc, float roughness)
 {
 }
 
-__device float3 bsdf_reflection_eval_reflect(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
+ccl_device float3 bsdf_reflection_eval_reflect(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
 {
 	return make_float3(0.0f, 0.0f, 0.0f);
 }
 
-__device float3 bsdf_reflection_eval_transmit(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
+ccl_device float3 bsdf_reflection_eval_transmit(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
 {
 	return make_float3(0.0f, 0.0f, 0.0f);
 }
 
-__device int bsdf_reflection_sample(const ShaderClosure *sc, float3 Ng, float3 I, float3 dIdx, float3 dIdy, float randu, float randv, float3 *eval, float3 *omega_in, float3 *domega_in_dx, float3 *domega_in_dy, float *pdf)
+ccl_device int bsdf_reflection_sample(const ShaderClosure *sc, float3 Ng, float3 I, float3 dIdx, float3 dIdy, float randu, float randv, float3 *eval, float3 *omega_in, float3 *domega_in_dx, float3 *domega_in_dy, float *pdf)
 {
 	float3 N = sc->N;
 
diff --git a/intern/cycles/kernel/closure/bsdf_refraction.h b/intern/cycles/kernel/closure/bsdf_refraction.h
index 8565c99d04e..c4698b42060 100644
--- a/intern/cycles/kernel/closure/bsdf_refraction.h
+++ b/intern/cycles/kernel/closure/bsdf_refraction.h
@@ -37,27 +37,27 @@ CCL_NAMESPACE_BEGIN
 
 /* REFRACTION */
 
-__device int bsdf_refraction_setup(ShaderClosure *sc)
+ccl_device int bsdf_refraction_setup(ShaderClosure *sc)
 {
 	sc->type = CLOSURE_BSDF_REFRACTION_ID;
 	return SD_BSDF;
 }
 
-__device void bsdf_refraction_blur(ShaderClosure *sc, float roughness)
+ccl_device void bsdf_refraction_blur(ShaderClosure *sc, float roughness)
 {
 }
 
-__device float3 bsdf_refraction_eval_reflect(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
+ccl_device float3 bsdf_refraction_eval_reflect(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
 {
 	return make_float3(0.0f, 0.0f, 0.0f);
 }
 
-__device float3 bsdf_refraction_eval_transmit(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
+ccl_device float3 bsdf_refraction_eval_transmit(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
 {
 	return make_float3(0.0f, 0.0f, 0.0f);
 }
 
-__device int bsdf_refraction_sample(const ShaderClosure *sc, float3 Ng, float3 I, float3 dIdx, float3 dIdy, float randu, float randv, float3 *eval, float3 *omega_in, float3 *domega_in_dx, float3 *domega_in_dy, float *pdf)
+ccl_device int bsdf_refraction_sample(const ShaderClosure *sc, float3 Ng, float3 I, float3 dIdx, float3 dIdy, float randu, float randv, float3 *eval, float3 *omega_in, float3 *domega_in_dx, float3 *domega_in_dy, float *pdf)
 {
 	float m_eta = sc->data0;
 	float3 N = sc->N;
diff --git a/intern/cycles/kernel/closure/bsdf_toon.h b/intern/cycles/kernel/closure/bsdf_toon.h
index e69981dba77..797fa4227ae 100644
--- a/intern/cycles/kernel/closure/bsdf_toon.h
+++ b/intern/cycles/kernel/closure/bsdf_toon.h
@@ -37,7 +37,7 @@ CCL_NAMESPACE_BEGIN
 
 /* DIFFUSE TOON */
 
-__device int bsdf_diffuse_toon_setup(ShaderClosure *sc)
+ccl_device int bsdf_diffuse_toon_setup(ShaderClosure *sc)
 {
 	sc->type = CLOSURE_BSDF_DIFFUSE_TOON_ID;
 	sc->data0 = clamp(sc->data0, 0.0f, 1.0f);
@@ -46,11 +46,11 @@ __device int bsdf_diffuse_toon_setup(ShaderClosure *sc)
 	return SD_BSDF|SD_BSDF_HAS_EVAL;
 }
 
-__device void bsdf_diffuse_toon_blur(ShaderClosure *sc, float roughness)
+ccl_device void bsdf_diffuse_toon_blur(ShaderClosure *sc, float roughness)
 {
 }
 
-__device float3 bsdf_toon_get_intensity(float max_angle, float smooth, float angle)
+ccl_device float3 bsdf_toon_get_intensity(float max_angle, float smooth, float angle)
 {
 	float is;
 
@@ -64,12 +64,12 @@ __device float3 bsdf_toon_get_intensity(float max_angle, float smooth, float ang
 	return make_float3(is, is, is);
 }
 
-__device float bsdf_toon_get_sample_angle(float max_angle, float smooth)
+ccl_device float bsdf_toon_get_sample_angle(float max_angle, float smooth)
 {
 	return fminf(max_angle + smooth, M_PI_2_F);
 }
 
-__device float3 bsdf_diffuse_toon_eval_reflect(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
+ccl_device float3 bsdf_diffuse_toon_eval_reflect(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
 {
 	float max_angle = sc->data0*M_PI_2_F;
 	float smooth = sc->data1*M_PI_2_F;
@@ -87,12 +87,12 @@ __device float3 bsdf_diffuse_toon_eval_reflect(const ShaderClosure *sc, const fl
 	return make_float3(0.0f, 0.0f, 0.0f);
 }
 
-__device float3 bsdf_diffuse_toon_eval_transmit(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
+ccl_device float3 bsdf_diffuse_toon_eval_transmit(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
 {
 	return make_float3(0.0f, 0.0f, 0.0f);
 }
 
-__device int bsdf_diffuse_toon_sample(const ShaderClosure *sc, float3 Ng, float3 I, float3 dIdx, float3 dIdy, float randu, float randv, float3 *eval, float3 *omega_in, float3 *domega_in_dx, float3 *domega_in_dy, float *pdf)
+ccl_device int bsdf_diffuse_toon_sample(const ShaderClosure *sc, float3 Ng, float3 I, float3 dIdx, float3 dIdy, float randu, float randv, float3 *eval, float3 *omega_in, float3 *domega_in_dx, float3 *domega_in_dy, float *pdf)
 {
 	float max_angle = sc->data0*M_PI_2_F;
 	float smooth = sc->data1*M_PI_2_F;
@@ -121,7 +121,7 @@ __device int bsdf_diffuse_toon_sample(const ShaderClosure *sc, float3 Ng, float3
 
 /* GLOSSY TOON */
 
-__device int bsdf_glossy_toon_setup(ShaderClosure *sc)
+ccl_device int bsdf_glossy_toon_setup(ShaderClosure *sc)
 {
 	sc->type = CLOSURE_BSDF_GLOSSY_TOON_ID;
 	sc->data0 = clamp(sc->data0, 0.0f, 1.0f);
@@ -130,11 +130,11 @@ __device int bsdf_glossy_toon_setup(ShaderClosure *sc)
 	return SD_BSDF|SD_BSDF_HAS_EVAL;
 }
 
-__device void bsdf_glossy_toon_blur(ShaderClosure *sc, float roughness)
+ccl_device void bsdf_glossy_toon_blur(ShaderClosure *sc, float roughness)
 {
 }
 
-__device float3 bsdf_glossy_toon_eval_reflect(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
+ccl_device float3 bsdf_glossy_toon_eval_reflect(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
 {
 	float max_angle = sc->data0*M_PI_2_F;
 	float smooth = sc->data1*M_PI_2_F;
@@ -158,12 +158,12 @@ __device float3 bsdf_glossy_toon_eval_reflect(const ShaderClosure *sc, const flo
 	return make_float3(0.0f, 0.0f, 0.0f);
 }
 
-__device float3 bsdf_glossy_toon_eval_transmit(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
+ccl_device float3 bsdf_glossy_toon_eval_transmit(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
 {
 	return make_float3(0.0f, 0.0f, 0.0f);
 }
 
-__device int bsdf_glossy_toon_sample(const ShaderClosure *sc, float3 Ng, float3 I, float3 dIdx, float3 dIdy, float randu, float randv, float3 *eval, float3 *omega_in, float3 *domega_in_dx, float3 *domega_in_dy, float *pdf)
+ccl_device int bsdf_glossy_toon_sample(const ShaderClosure *sc, float3 Ng, float3 I, float3 dIdx, float3 dIdy, float randu, float randv, float3 *eval, float3 *omega_in, float3 *domega_in_dx, float3 *domega_in_dy, float *pdf)
 {
 	float max_angle = sc->data0*M_PI_2_F;
 	float smooth = sc->data1*M_PI_2_F;
diff --git a/intern/cycles/kernel/closure/bsdf_transparent.h b/intern/cycles/kernel/closure/bsdf_transparent.h
index 81bc7690b50..e62aecf3da6 100644
--- a/intern/cycles/kernel/closure/bsdf_transparent.h
+++ b/intern/cycles/kernel/closure/bsdf_transparent.h
@@ -35,27 +35,27 @@
 
 CCL_NAMESPACE_BEGIN
 
-__device int bsdf_transparent_setup(ShaderClosure *sc)
+ccl_device int bsdf_transparent_setup(ShaderClosure *sc)
 {
 	sc->type = CLOSURE_BSDF_TRANSPARENT_ID;
 	return SD_BSDF;
 }
 
-__device void bsdf_transparent_blur(ShaderClosure *sc, float roughness)
+ccl_device void bsdf_transparent_blur(ShaderClosure *sc, float roughness)
 {
 }
 
-__device float3 bsdf_transparent_eval_reflect(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
+ccl_device float3 bsdf_transparent_eval_reflect(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
 {
 	return make_float3(0.0f, 0.0f, 0.0f);
 }
 
-__device float3 bsdf_transparent_eval_transmit(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
+ccl_device float3 bsdf_transparent_eval_transmit(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
 {
 	return make_float3(0.0f, 0.0f, 0.0f);
 }
 
-__device int bsdf_transparent_sample(const ShaderClosure *sc, float3 Ng, float3 I, float3 dIdx, float3 dIdy, float randu, float randv, float3 *eval, float3 *omega_in, float3 *domega_in_dx, float3 *domega_in_dy, float *pdf)
+ccl_device int bsdf_transparent_sample(const ShaderClosure *sc, float3 Ng, float3 I, float3 dIdx, float3 dIdy, float randu, float randv, float3 *eval, float3 *omega_in, float3 *domega_in_dx, float3 *domega_in_dy, float *pdf)
 {
 	// only one direction is possible
 	*omega_in = -I;
diff --git a/intern/cycles/kernel/closure/bsdf_util.h b/intern/cycles/kernel/closure/bsdf_util.h
index 5ac26055e8d..f6dceb3ca82 100644
--- a/intern/cycles/kernel/closure/bsdf_util.h
+++ b/intern/cycles/kernel/closure/bsdf_util.h
@@ -35,7 +35,7 @@
 
 CCL_NAMESPACE_BEGIN
 
-__device float fresnel_dielectric(float eta, const float3 N,
+ccl_device float fresnel_dielectric(float eta, const float3 N,
 		const float3 I, float3 *R, float3 *T,
 #ifdef __RAY_DIFFERENTIALS__
 		const float3 dIdx, const float3 dIdy,
@@ -95,7 +95,7 @@ __device float fresnel_dielectric(float eta, const float3 N,
 	}
 }
 
-__device float fresnel_dielectric_cos(float cosi, float eta)
+ccl_device float fresnel_dielectric_cos(float cosi, float eta)
 {
 	// compute fresnel reflectance without explicitly computing
 	// the refracted direction
@@ -110,7 +110,7 @@ __device float fresnel_dielectric_cos(float cosi, float eta)
 	return 1.0f; // TIR(no refracted component)
 }
 
-__device float fresnel_conductor(float cosi, float eta, float k)
+ccl_device float fresnel_conductor(float cosi, float eta, float k)
 {
 	float tmp_f = eta * eta + k * k;
 	float tmp = tmp_f * cosi * cosi;
@@ -121,7 +121,7 @@ __device float fresnel_conductor(float cosi, float eta, float k)
 	return(Rparl2 + Rperp2) * 0.5f;
 }
 
-__device float smooth_step(float edge0, float edge1, float x)
+ccl_device float smooth_step(float edge0, float edge1, float x)
 {
 	float result;
 	if(x < edge0) result = 0.0f;
diff --git a/intern/cycles/kernel/closure/bsdf_ward.h b/intern/cycles/kernel/closure/bsdf_ward.h
index 0e5b0c544c7..c9de615a011 100644
--- a/intern/cycles/kernel/closure/bsdf_ward.h
+++ b/intern/cycles/kernel/closure/bsdf_ward.h
@@ -37,7 +37,7 @@ CCL_NAMESPACE_BEGIN
 
 /* WARD */
 
-__device int bsdf_ward_setup(ShaderClosure *sc)
+ccl_device int bsdf_ward_setup(ShaderClosure *sc)
 {
 	sc->data0 = clamp(sc->data0, 1e-4f, 1.0f); /* m_ax */
 	sc->data1 = clamp(sc->data1, 1e-4f, 1.0f); /* m_ay */
@@ -46,13 +46,13 @@ __device int bsdf_ward_setup(ShaderClosure *sc)
 	return SD_BSDF|SD_BSDF_HAS_EVAL|SD_BSDF_GLOSSY;
 }
 
-__device void bsdf_ward_blur(ShaderClosure *sc, float roughness)
+ccl_device void bsdf_ward_blur(ShaderClosure *sc, float roughness)
 {
 	sc->data0 = fmaxf(roughness, sc->data0); /* m_ax */
 	sc->data1 = fmaxf(roughness, sc->data1); /* m_ay */
 }
 
-__device float3 bsdf_ward_eval_reflect(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
+ccl_device float3 bsdf_ward_eval_reflect(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
 {
 	float m_ax = sc->data0;
 	float m_ay = sc->data1;
@@ -87,12 +87,12 @@ __device float3 bsdf_ward_eval_reflect(const ShaderClosure *sc, const float3 I,
 	return make_float3 (0, 0, 0);
 }
 
-__device float3 bsdf_ward_eval_transmit(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
+ccl_device float3 bsdf_ward_eval_transmit(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
 {
 	return make_float3(0.0f, 0.0f, 0.0f);
 }
 
-__device int bsdf_ward_sample(const ShaderClosure *sc, float3 Ng, float3 I, float3 dIdx, float3 dIdy, float randu, float randv, float3 *eval, float3 *omega_in, float3 *domega_in_dx, float3 *domega_in_dy, float *pdf)
+ccl_device int bsdf_ward_sample(const ShaderClosure *sc, float3 Ng, float3 I, float3 dIdx, float3 dIdy, float randu, float randv, float3 *eval, float3 *omega_in, float3 *domega_in_dx, float3 *domega_in_dy, float *pdf)
 {
 	float m_ax = sc->data0;
 	float m_ay = sc->data1;
diff --git a/intern/cycles/kernel/closure/bsdf_westin.h b/intern/cycles/kernel/closure/bsdf_westin.h
index e1a6b031d5e..ca4c05e91fe 100644
--- a/intern/cycles/kernel/closure/bsdf_westin.h
+++ b/intern/cycles/kernel/closure/bsdf_westin.h
@@ -37,7 +37,7 @@ CCL_NAMESPACE_BEGIN
 
 /* WESTIN BACKSCATTER */
 
-__device int bsdf_westin_backscatter_setup(ShaderClosure *sc)
+ccl_device int bsdf_westin_backscatter_setup(ShaderClosure *sc)
 {
 	float roughness = sc->data0;
 	roughness = clamp(roughness, 1e-5f, 1.0f);
@@ -49,14 +49,14 @@ __device int bsdf_westin_backscatter_setup(ShaderClosure *sc)
 	return SD_BSDF|SD_BSDF_HAS_EVAL|SD_BSDF_GLOSSY;
 }
 
-__device void bsdf_westin_backscatter_blur(ShaderClosure *sc, float roughness)
+ccl_device void bsdf_westin_backscatter_blur(ShaderClosure *sc, float roughness)
 {
 	float m_invroughness = sc->data0;
 	m_invroughness = min(1.0f/roughness, m_invroughness);
 	sc->data0 = m_invroughness;
 }
 
-__device float3 bsdf_westin_backscatter_eval_reflect(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
+ccl_device float3 bsdf_westin_backscatter_eval_reflect(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
 {
 	float m_invroughness = sc->data0;
 	float3 N = sc->N;
@@ -73,12 +73,12 @@ __device float3 bsdf_westin_backscatter_eval_reflect(const ShaderClosure *sc, co
 	return make_float3 (0, 0, 0);
 }
 
-__device float3 bsdf_westin_backscatter_eval_transmit(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
+ccl_device float3 bsdf_westin_backscatter_eval_transmit(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
 {
 	return make_float3(0.0f, 0.0f, 0.0f);
 }
 
-__device int bsdf_westin_backscatter_sample(const ShaderClosure *sc, float3 Ng, float3 I, float3 dIdx, float3 dIdy, float randu, float randv, float3 *eval, float3 *omega_in, float3 *domega_in_dx, float3 *domega_in_dy, float *pdf)
+ccl_device int bsdf_westin_backscatter_sample(const ShaderClosure *sc, float3 Ng, float3 I, float3 dIdx, float3 dIdy, float randu, float randv, float3 *eval, float3 *omega_in, float3 *domega_in_dx, float3 *domega_in_dy, float *pdf)
 {
 	float m_invroughness = sc->data0;
 	float3 N = sc->N;
@@ -116,18 +116,18 @@ __device int bsdf_westin_backscatter_sample(const ShaderClosure *sc, float3 Ng,
 
 /* WESTIN SHEEN */
 
-__device int bsdf_westin_sheen_setup(ShaderClosure *sc)
+ccl_device int bsdf_westin_sheen_setup(ShaderClosure *sc)
 {
 	/* float edginess = sc->data0; */
 	sc->type = CLOSURE_BSDF_WESTIN_SHEEN_ID;
 	return SD_BSDF|SD_BSDF_HAS_EVAL|SD_BSDF_GLOSSY;
 }
 
-__device void bsdf_westin_sheen_blur(ShaderClosure *sc, float roughness)
+ccl_device void bsdf_westin_sheen_blur(ShaderClosure *sc, float roughness)
 {
 }
 
-__device float3 bsdf_westin_sheen_eval_reflect(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
+ccl_device float3 bsdf_westin_sheen_eval_reflect(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
 {
 	float m_edginess = sc->data0;
 	float3 N = sc->N;
@@ -144,12 +144,12 @@ __device float3 bsdf_westin_sheen_eval_reflect(const ShaderClosure *sc, const fl
 	return make_float3 (0, 0, 0);
 }
 
-__device float3 bsdf_westin_sheen_eval_transmit(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
+ccl_device float3 bsdf_westin_sheen_eval_transmit(const ShaderClosure *sc, const float3 I, const float3 omega_in, float *pdf)
 {
 	return make_float3(0.0f, 0.0f, 0.0f);
 }
 
-__device int bsdf_westin_sheen_sample(const ShaderClosure *sc, float3 Ng, float3 I, float3 dIdx, float3 dIdy, float randu, float randv, float3 *eval, float3 *omega_in, float3 *domega_in_dx, float3 *domega_in_dy, float *pdf)
+ccl_device int bsdf_westin_sheen_sample(const ShaderClosure *sc, float3 Ng, float3 I, float3 dIdx, float3 dIdy, float randu, float randv, float3 *eval, float3 *omega_in, float3 *domega_in_dx, float3 *domega_in_dy, float *pdf)
 {
 	float m_edginess = sc->data0;
 	float3 N = sc->N;
diff --git a/intern/cycles/kernel/closure/bssrdf.h b/intern/cycles/kernel/closure/bssrdf.h
index 4ceff655dd5..3849dedc3b6 100644
--- a/intern/cycles/kernel/closure/bssrdf.h
+++ b/intern/cycles/kernel/closure/bssrdf.h
@@ -19,7 +19,7 @@
 
 CCL_NAMESPACE_BEGIN
 
-__device int bssrdf_setup(ShaderClosure *sc, ClosureType type)
+ccl_device int bssrdf_setup(ShaderClosure *sc, ClosureType type)
 {
 	if(sc->data0 < BSSRDF_MIN_RADIUS) {
 		/* revert to diffuse BSDF if radius too small */
@@ -47,7 +47,7 @@ __device int bssrdf_setup(ShaderClosure *sc, ClosureType type)
 /* paper suggests 1/12.46 which is much too small, suspect it's *12.46 */
 #define GAUSS_TRUNCATE 12.46f
 
-__device float bssrdf_gaussian_eval(ShaderClosure *sc, float r)
+ccl_device float bssrdf_gaussian_eval(ShaderClosure *sc, float r)
 {
 	/* integrate (2*pi*r * exp(-r*r/(2*v)))/(2*pi*v)) from 0 to Rm
 	 * = 1 - exp(-Rm*Rm/(2*v)) */
@@ -60,7 +60,7 @@ __device float bssrdf_gaussian_eval(ShaderClosure *sc, float r)
 	return expf(-r*r/(2.0f*v))/(2.0f*M_PI_F*v);
 }
 
-__device float bssrdf_gaussian_pdf(ShaderClosure *sc, float r)
+ccl_device float bssrdf_gaussian_pdf(ShaderClosure *sc, float r)
 {
 	/* 1.0 - expf(-Rm*Rm/(2*v)) simplified */
 	const float area_truncated = 1.0f - expf(-0.5f*GAUSS_TRUNCATE);
@@ -68,7 +68,7 @@ __device float bssrdf_gaussian_pdf(ShaderClosure *sc, float r)
 	return bssrdf_gaussian_eval(sc, r) * (1.0f/(area_truncated));
 }
 
-__device void bssrdf_gaussian_sample(ShaderClosure *sc, float xi, float *r, float *h)
+ccl_device void bssrdf_gaussian_sample(ShaderClosure *sc, float xi, float *r, float *h)
 {
 	/* xi = integrate (2*pi*r * exp(-r*r/(2*v)))/(2*pi*v)) = -exp(-r^2/(2*v))
 	 * r = sqrt(-2*v*logf(xi)) */
@@ -94,7 +94,7 @@ __device void bssrdf_gaussian_sample(ShaderClosure *sc, float xi, float *r, floa
  * far as I can tell has no closed form solution. So we get an iterative solution
  * instead with newton-raphson. */
 
-__device float bssrdf_cubic_eval(ShaderClosure *sc, float r)
+ccl_device float bssrdf_cubic_eval(ShaderClosure *sc, float r)
 {
 	const float sharpness = sc->T.x;
 
@@ -141,13 +141,13 @@ __device float bssrdf_cubic_eval(ShaderClosure *sc, float r)
 	}
 }
 
-__device float bssrdf_cubic_pdf(ShaderClosure *sc, float r)
+ccl_device float bssrdf_cubic_pdf(ShaderClosure *sc, float r)
 {
 	return bssrdf_cubic_eval(sc, r);
 }
 
 /* solve 10x^2 - 20x^3 + 15x^4 - 4x^5 - xi == 0 */
-__device float bssrdf_cubic_quintic_root_find(float xi)
+ccl_device float bssrdf_cubic_quintic_root_find(float xi)
 {
 	/* newton-raphson iteration, usually succeeds in 2-4 iterations, except
 	 * outside 0.02 ... 0.98 where it can go up to 10, so overall performance
@@ -174,7 +174,7 @@ __device float bssrdf_cubic_quintic_root_find(float xi)
 	return x;
 }
 
-__device void bssrdf_cubic_sample(ShaderClosure *sc, float xi, float *r, float *h)
+ccl_device void bssrdf_cubic_sample(ShaderClosure *sc, float xi, float *r, float *h)
 {
 	float Rm = sc->data0;
 	float r_ = bssrdf_cubic_quintic_root_find(xi);
@@ -196,13 +196,13 @@ __device void bssrdf_cubic_sample(ShaderClosure *sc, float xi, float *r, float *
  * 
  * Samples distributed over disk with no falloff, for reference. */
 
-__device float bssrdf_none_eval(ShaderClosure *sc, float r)
+ccl_device float bssrdf_none_eval(ShaderClosure *sc, float r)
 {
 	const float Rm = sc->data0;
 	return (r < Rm)? 1.0f: 0.0f;
 }
 
-__device float bssrdf_none_pdf(ShaderClosure *sc, float r)
+ccl_device float bssrdf_none_pdf(ShaderClosure *sc, float r)
 {
 	/* integrate (2*pi*r)/(pi*Rm*Rm) from 0 to Rm = 1 */
 	const float Rm = sc->data0;
@@ -211,7 +211,7 @@ __device float bssrdf_none_pdf(ShaderClosure *sc, float r)
 	return bssrdf_none_eval(sc, r) / area;
 }
 
-__device void bssrdf_none_sample(ShaderClosure *sc, float xi, float *r, float *h)
+ccl_device void bssrdf_none_sample(ShaderClosure *sc, float xi, float *r, float *h)
 {
 	/* xi = integrate (2*pi*r)/(pi*Rm*Rm) = r^2/Rm^2
 	 * r = sqrt(xi)*Rm */
@@ -226,7 +226,7 @@ __device void bssrdf_none_sample(ShaderClosure *sc, float xi, float *r, float *h
 
 /* Generic */
 
-__device void bssrdf_sample(ShaderClosure *sc, float xi, float *r, float *h)
+ccl_device void bssrdf_sample(ShaderClosure *sc, float xi, float *r, float *h)
 {
 	if(sc->type == CLOSURE_BSSRDF_CUBIC_ID)
 		bssrdf_cubic_sample(sc, xi, r, h);
@@ -234,7 +234,7 @@ __device void bssrdf_sample(ShaderClosure *sc, float xi, float *r, float *h)
 		bssrdf_gaussian_sample(sc, xi, r, h);
 }
 
-__device float bssrdf_pdf(ShaderClosure *sc, float r)
+ccl_device float bssrdf_pdf(ShaderClosure *sc, float r)
 {
 	if(sc->type == CLOSURE_BSSRDF_CUBIC_ID)
 		return bssrdf_cubic_pdf(sc, r);
diff --git a/intern/cycles/kernel/closure/emissive.h b/intern/cycles/kernel/closure/emissive.h
index 33b1b695a9a..c534df373bd 100644
--- a/intern/cycles/kernel/closure/emissive.h
+++ b/intern/cycles/kernel/closure/emissive.h
@@ -37,19 +37,19 @@ CCL_NAMESPACE_BEGIN
 /* return the probability distribution function in the direction I,
  * given the parameters and the light's surface normal.  This MUST match
  * the PDF computed by sample(). */
-__device float emissive_pdf(const float3 Ng, const float3 I)
+ccl_device float emissive_pdf(const float3 Ng, const float3 I)
 {
 	float cosNO = fabsf(dot(Ng, I));
 	return (cosNO > 0.0f)? 1.0f: 0.0f;
 }
 
-__device void emissive_sample(const float3 Ng, float randu, float randv,
+ccl_device void emissive_sample(const float3 Ng, float randu, float randv,
 	float3 *omega_out, float *pdf)
 {
 	/* todo: not implemented and used yet */
 }
 
-__device float3 emissive_simple_eval(const float3 Ng, const float3 I)
+ccl_device float3 emissive_simple_eval(const float3 Ng, const float3 I)
 {
 	float res = emissive_pdf(Ng, I);
 	
diff --git a/intern/cycles/kernel/closure/volume.h b/intern/cycles/kernel/closure/volume.h
index ddaf939984e..f30b30c8c76 100644
--- a/intern/cycles/kernel/closure/volume.h
+++ b/intern/cycles/kernel/closure/volume.h
@@ -21,7 +21,7 @@ CCL_NAMESPACE_BEGIN
 
 /* ISOTROPIC VOLUME CLOSURE */
 
-__device int volume_isotropic_setup(ShaderClosure *sc, float density)
+ccl_device int volume_isotropic_setup(ShaderClosure *sc, float density)
 {
 	sc->type = CLOSURE_VOLUME_ISOTROPIC_ID;
 	sc->data0 = density;
@@ -29,14 +29,14 @@ __device int volume_isotropic_setup(ShaderClosure *sc, float density)
 	return SD_VOLUME;
 }
 
-__device float3 volume_isotropic_eval_phase(const ShaderClosure *sc, const float3 omega_in, const float3 omega_out)
+ccl_device float3 volume_isotropic_eval_phase(const ShaderClosure *sc, const float3 omega_in, const float3 omega_out)
 {
 	return make_float3(1.0f, 1.0f, 1.0f);
 }
 
 /* TRANSPARENT VOLUME CLOSURE */
 
-__device int volume_transparent_setup(ShaderClosure *sc, float density)
+ccl_device int volume_transparent_setup(ShaderClosure *sc, float density)
 {
 	sc->type = CLOSURE_VOLUME_TRANSPARENT_ID;
 	sc->data0 = density;
@@ -44,14 +44,14 @@ __device int volume_transparent_setup(ShaderClosure *sc, float density)
 	return SD_VOLUME;
 }
 
-__device float3 volume_transparent_eval_phase(const ShaderClosure *sc, const float3 omega_in, const float3 omega_out)
+ccl_device float3 volume_transparent_eval_phase(const ShaderClosure *sc, const float3 omega_in, const float3 omega_out)
 {
 	return make_float3(1.0f, 1.0f, 1.0f);
 }
 
 /* VOLUME CLOSURE */
 
-__device float3 volume_eval_phase(KernelGlobals *kg, const ShaderClosure *sc, const float3 omega_in, const float3 omega_out)
+ccl_device float3 volume_eval_phase(KernelGlobals *kg, const ShaderClosure *sc, const float3 omega_in, const float3 omega_out)
 {
 #ifdef __OSL__
 	if(kg->osl && sc->prim)