diff --git a/drivers/gpu/nvgpu/common/vgpu/gp10b/vgpu_hal_gp10b.c b/drivers/gpu/nvgpu/common/vgpu/gp10b/vgpu_hal_gp10b.c
index 023ee9819..d26dc9a11 100644
--- a/drivers/gpu/nvgpu/common/vgpu/gp10b/vgpu_hal_gp10b.c
+++ b/drivers/gpu/nvgpu/common/vgpu/gp10b/vgpu_hal_gp10b.c
@@ -292,7 +292,8 @@ static const struct gpu_ops vgpu_gp10b_ops = {
 		.fecs_trace = {
 			.alloc_user_buffer = vgpu_alloc_user_buffer,
 			.free_user_buffer = vgpu_free_user_buffer,
-			.mmap_user_buffer = vgpu_mmap_user_buffer,
+			.get_mmap_user_buffer_info =
+				vgpu_get_mmap_user_buffer_info,
 			.init = vgpu_fecs_trace_init,
 			.deinit = vgpu_fecs_trace_deinit,
 			.enable = vgpu_fecs_trace_enable,
diff --git a/drivers/gpu/nvgpu/common/vgpu/gr/fecs_trace_vgpu.h b/drivers/gpu/nvgpu/common/vgpu/gr/fecs_trace_vgpu.h
index 30b9721a1..734bb6eed 100644
--- a/drivers/gpu/nvgpu/common/vgpu/gr/fecs_trace_vgpu.h
+++ b/drivers/gpu/nvgpu/common/vgpu/gr/fecs_trace_vgpu.h
@@ -50,7 +50,8 @@ bool vgpu_fecs_trace_is_enabled(struct gk20a *g);
 int vgpu_fecs_trace_poll(struct gk20a *g);
 int vgpu_alloc_user_buffer(struct gk20a *g, void **buf, size_t *size);
 int vgpu_free_user_buffer(struct gk20a *g);
-int vgpu_mmap_user_buffer(struct gk20a *g, struct vm_area_struct *vma);
+void vgpu_get_mmap_user_buffer_info(struct gk20a *g,
+				void **mmapaddr, size_t *mmapsize);
 int vgpu_fecs_trace_max_entries(struct gk20a *g,
 			struct nvgpu_gpu_ctxsw_trace_filter *filter);
 int vgpu_fecs_trace_set_filter(struct gk20a *g,
diff --git a/drivers/gpu/nvgpu/common/vgpu/gv11b/vgpu_hal_gv11b.c b/drivers/gpu/nvgpu/common/vgpu/gv11b/vgpu_hal_gv11b.c
index 022706e4b..ac99b9d8b 100644
--- a/drivers/gpu/nvgpu/common/vgpu/gv11b/vgpu_hal_gv11b.c
+++ b/drivers/gpu/nvgpu/common/vgpu/gv11b/vgpu_hal_gv11b.c
@@ -341,7 +341,8 @@ static const struct gpu_ops vgpu_gv11b_ops = {
 		.fecs_trace = {
 			.alloc_user_buffer = vgpu_alloc_user_buffer,
 			.free_user_buffer = vgpu_free_user_buffer,
-			.mmap_user_buffer = vgpu_mmap_user_buffer,
+			.get_mmap_user_buffer_info =
+				vgpu_get_mmap_user_buffer_info,
 			.init = vgpu_fecs_trace_init,
 			.deinit = vgpu_fecs_trace_deinit,
 			.enable = vgpu_fecs_trace_enable,
diff --git a/drivers/gpu/nvgpu/gp10b/hal_gp10b.c b/drivers/gpu/nvgpu/gp10b/hal_gp10b.c
index 3101777b8..c1fd40144 100644
--- a/drivers/gpu/nvgpu/gp10b/hal_gp10b.c
+++ b/drivers/gpu/nvgpu/gp10b/hal_gp10b.c
@@ -443,7 +443,8 @@ static const struct gpu_ops gp10b_ops = {
 		.fecs_trace = {
 			.alloc_user_buffer = nvgpu_gr_fecs_trace_ring_alloc,
 			.free_user_buffer = nvgpu_gr_fecs_trace_ring_free,
-			.mmap_user_buffer = nvgpu_gr_fecs_trace_mmap_buffer,
+			.get_mmap_user_buffer_info =
+				nvgpu_gr_fecs_trace_get_mmap_buffer_info,
 			.init = nvgpu_gr_fecs_trace_init,
 			.deinit = nvgpu_gr_fecs_trace_deinit,
 			.enable = nvgpu_gr_fecs_trace_enable,
diff --git a/drivers/gpu/nvgpu/gv100/hal_gv100.c b/drivers/gpu/nvgpu/gv100/hal_gv100.c
index 5c32e8eac..7262b83bb 100644
--- a/drivers/gpu/nvgpu/gv100/hal_gv100.c
+++ b/drivers/gpu/nvgpu/gv100/hal_gv100.c
@@ -561,7 +561,8 @@ static const struct gpu_ops gv100_ops = {
 		.fecs_trace = {
 			.alloc_user_buffer = nvgpu_gr_fecs_trace_ring_alloc,
 			.free_user_buffer = nvgpu_gr_fecs_trace_ring_free,
-			.mmap_user_buffer = nvgpu_gr_fecs_trace_mmap_buffer,
+			.get_mmap_user_buffer_info =
+				nvgpu_gr_fecs_trace_get_mmap_buffer_info,
 			.init = nvgpu_gr_fecs_trace_init,
 			.deinit = nvgpu_gr_fecs_trace_deinit,
 			.enable = nvgpu_gr_fecs_trace_enable,
diff --git a/drivers/gpu/nvgpu/gv11b/hal_gv11b.c b/drivers/gpu/nvgpu/gv11b/hal_gv11b.c
index 8c7341cc5..8b25acde1 100644
--- a/drivers/gpu/nvgpu/gv11b/hal_gv11b.c
+++ b/drivers/gpu/nvgpu/gv11b/hal_gv11b.c
@@ -540,7 +540,8 @@ static const struct gpu_ops gv11b_ops = {
 		.fecs_trace = {
 			.alloc_user_buffer = nvgpu_gr_fecs_trace_ring_alloc,
 			.free_user_buffer = nvgpu_gr_fecs_trace_ring_free,
-			.mmap_user_buffer = nvgpu_gr_fecs_trace_mmap_buffer,
+			.get_mmap_user_buffer_info =
+				nvgpu_gr_fecs_trace_get_mmap_buffer_info,
 			.init = nvgpu_gr_fecs_trace_init,
 			.deinit = nvgpu_gr_fecs_trace_deinit,
 			.enable = nvgpu_gr_fecs_trace_enable,
diff --git a/drivers/gpu/nvgpu/include/nvgpu/gk20a.h b/drivers/gpu/nvgpu/include/nvgpu/gk20a.h
index 34323029c..fd3588e69 100644
--- a/drivers/gpu/nvgpu/include/nvgpu/gk20a.h
+++ b/drivers/gpu/nvgpu/include/nvgpu/gk20a.h
@@ -173,7 +173,7 @@ struct railgate_stats {
 #define GPU_LIT_PERFMON_PMMFBP_ROP_DOMAIN_START 43
 #define GPU_LIT_PERFMON_PMMFBP_ROP_DOMAIN_COUNT 44
 
-#define nvgpu_get_litter_value(g, v) (g)->ops.get_litter_value((g), v)
+#define nvgpu_get_litter_value(g, v) ((g)->ops.get_litter_value((g), v))
 
 #define MAX_TPC_PG_CONFIGS      3
 
@@ -274,7 +274,7 @@ struct gpu_ops {
 		int (*handle_sw_method)(struct gk20a *g, u32 addr,
 					 u32 class_num, u32 offset, u32 data);
 		void (*set_alpha_circular_buffer_size)(struct gk20a *g,
-					               u32 data);
+							u32 data);
 		void (*set_circular_buffer_size)(struct gk20a *g, u32 data);
 		void (*set_bes_crop_debug3)(struct gk20a *g, u32 data);
 		void (*set_bes_crop_debug4)(struct gk20a *g, u32 data);
@@ -612,8 +612,8 @@ struct gpu_ops {
 			int (*alloc_user_buffer)(struct gk20a *g,
 						void **buf, size_t *size);
 			int (*free_user_buffer)(struct gk20a *g);
-			int (*mmap_user_buffer)(struct gk20a *g,
-						struct vm_area_struct *vma);
+			void (*get_mmap_user_buffer_info)(struct gk20a *g,
+						void **addr, size_t *size);
 			int (*set_filter)(struct gk20a *g,
 				struct nvgpu_gpu_ctxsw_trace_filter *filter);
 			u32 (*get_buffer_full_mailbox_val)(void);
@@ -1333,7 +1333,7 @@ struct gpu_ops {
 		} boardobj;
 		struct {
 			int (*clk_set_boot_clk)(struct gk20a *g);
-		}clk;
+		} clk;
 	} pmu_ver;
 	struct {
 		int (*get_netlist_name)(struct gk20a *g, int index, char *name);
@@ -1566,7 +1566,7 @@ struct gpu_ops {
 		int (*arbiter_clk_init)(struct gk20a *g);
 		bool (*check_clk_arb_support)(struct gk20a *g);
 		u32 (*get_arbiter_clk_domains)(struct gk20a *g);
-		int (*get_arbiter_f_points)(struct gk20a *g,u32 api_domain,
+		int (*get_arbiter_f_points)(struct gk20a *g, u32 api_domain,
 				u32 *num_points, u16 *freqs_in_mhz);
 		int (*get_arbiter_clk_range)(struct gk20a *g, u32 api_domain,
 				u16 *min_mhz, u16 *max_mhz);
diff --git a/drivers/gpu/nvgpu/include/nvgpu/gr/fecs_trace.h b/drivers/gpu/nvgpu/include/nvgpu/gr/fecs_trace.h
index 6da7c8d28..9a5990257 100644
--- a/drivers/gpu/nvgpu/include/nvgpu/gr/fecs_trace.h
+++ b/drivers/gpu/nvgpu/include/nvgpu/gr/fecs_trace.h
@@ -176,7 +176,8 @@ int nvgpu_gr_fecs_trace_unbind_channel(struct gk20a *g,
  */
 int nvgpu_gr_fecs_trace_ring_alloc(struct gk20a *g, void **buf, size_t *size);
 int nvgpu_gr_fecs_trace_ring_free(struct gk20a *g);
-int nvgpu_gr_fecs_trace_mmap_buffer(struct gk20a *g, struct vm_area_struct *vma);
+void nvgpu_gr_fecs_trace_get_mmap_buffer_info(struct gk20a *g,
+				void **mmapaddr, size_t *mmapsize);
 void nvgpu_gr_fecs_trace_add_tsg_reset(struct gk20a *g, struct tsg_gk20a *tsg);
 u8 nvgpu_gpu_ctxsw_tags_to_common_tags(u8 tags);
 int nvgpu_gr_fecs_trace_write_entry(struct gk20a *g,
diff --git a/drivers/gpu/nvgpu/include/nvgpu/kref.h b/drivers/gpu/nvgpu/include/nvgpu/kref.h
index 460eb4b38..0e096deed 100644
--- a/drivers/gpu/nvgpu/include/nvgpu/kref.h
+++ b/drivers/gpu/nvgpu/include/nvgpu/kref.h
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2017-2018, NVIDIA CORPORATION.  All rights reserved.
+ * Copyright (c) 2017-2019, NVIDIA CORPORATION.  All rights reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -27,6 +27,7 @@
 #define NVGPU_KREF_H
 
 #include <nvgpu/atomic.h>
+#include <nvgpu/types.h>
 
 struct nvgpu_ref {
 	nvgpu_atomic_t refcount;
diff --git a/drivers/gpu/nvgpu/include/nvgpu/log2.h b/drivers/gpu/nvgpu/include/nvgpu/log2.h
index 98db1ecc7..f1f78d80f 100644
--- a/drivers/gpu/nvgpu/include/nvgpu/log2.h
+++ b/drivers/gpu/nvgpu/include/nvgpu/log2.h
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2017-2018, NVIDIA CORPORATION.  All rights reserved.
+ * Copyright (c) 2017-2019, NVIDIA CORPORATION.  All rights reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -24,7 +24,7 @@
 
 #ifdef __KERNEL__
 #include <linux/log2.h>
-#elif defined(__NVGPU_POSIX__)
+#else
 #include <nvgpu/posix/log2.h>
 #endif
 
diff --git a/drivers/gpu/nvgpu/include/nvgpu/posix/log2.h b/drivers/gpu/nvgpu/include/nvgpu/posix/log2.h
index e98768ab6..37420b35b 100644
--- a/drivers/gpu/nvgpu/include/nvgpu/posix/log2.h
+++ b/drivers/gpu/nvgpu/include/nvgpu/posix/log2.h
@@ -33,7 +33,6 @@
 #define is_power_of_2(x)					\
 	({							\
 		typeof(x) __x__ = (x);				\
-		(__x__ != 0U && ((__x__ & (__x__ - 1U)) == 0U));	\
+		(__x__ != 0U && ((__x__ & (__x__ - 1U)) == 0U));\
 	})
-
 #endif /* NVGPU_POSIX_LOG2_H */
diff --git a/drivers/gpu/nvgpu/include/nvgpu/posix/types.h b/drivers/gpu/nvgpu/include/nvgpu/posix/types.h
index 5550f88c8..488f554cd 100644
--- a/drivers/gpu/nvgpu/include/nvgpu/posix/types.h
+++ b/drivers/gpu/nvgpu/include/nvgpu/posix/types.h
@@ -91,16 +91,22 @@ typedef signed long long	s64;
 
 #define DIV_ROUND_UP(n, d)	(((n) + (d) - 1U) / (d))
 
-/*
- * Only used in clk_gm20b.c which we will never unit test. Don't use!
- */
-#define DIV_ROUND_CLOSEST(x, divisor)	({BUG(); 0; })
-
+#define DIV_ROUND_CLOSEST(a, divisor)(                  \
+{                                                       \
+	typeof(a) val = a;                              \
+	typeof(divisor) div = divisor;                  \
+	(((typeof(a))-1) > 0 ||                         \
+	((typeof(divisor))-1) > 0 || (val) > 0) ?       \
+		(((val) + ((div) / 2)) / (div)) :       \
+		(((val) - ((div) / 2)) / (div));        \
+}                                                       \
+)
 /*
  * Joys of userspace: usually division just works since the compiler can link
  * against external division functions implicitly.
  */
 #define do_div(a, b)		((a) /= (b))
+
 #define div64_u64(a, b)		((a) / (b))
 
 #define __round_mask(x, y)	((__typeof__(x))((y) - 1U))
@@ -177,12 +183,12 @@ static inline unsigned long __hweight64(uint64_t x)
  * Better suited under a compiler.h type header file, but for now these can live
  * here.
  */
-#define __must_check
+#define __must_check		__attribute__((warn_unused_result))
 #define __maybe_unused		__attribute__((unused))
 #define __iomem
 #define __user
-#define unlikely
-#define likely
+#define unlikely(x)	(x)
+#define likely(x)	(x)
 
 #define __stringify(x)		#x
 
@@ -196,9 +202,9 @@ static inline unsigned long __hweight64(uint64_t x)
 		*__p__ = (v);				\
 	})
 
-#define container_of(ptr, type, member) ({                      \
-        const typeof( ((type *)0)->member ) *__mptr = (ptr);    \
-        (type *)( (char *)__mptr - offsetof(type,member) );})
+#define container_of(ptr, type, member) ({                     \
+	const typeof( ((type *)0)->member ) *__mptr = (ptr);    \
+	(type *)( (char *)__mptr - offsetof(type,member) );})
 
 #define __packed __attribute__((packed))
 
diff --git a/drivers/gpu/nvgpu/include/nvgpu/types.h b/drivers/gpu/nvgpu/include/nvgpu/types.h
index 016b4ef22..670942f02 100644
--- a/drivers/gpu/nvgpu/include/nvgpu/types.h
+++ b/drivers/gpu/nvgpu/include/nvgpu/types.h
@@ -24,10 +24,8 @@
 
 #ifdef __KERNEL__
 #include <linux/types.h>
-#elif defined(__NVGPU_POSIX__)
-#include <nvgpu/posix/types.h>
 #else
-#include <nvgpu_rmos/include/types.h>
+#include <nvgpu/posix/types.h>
 #endif
 
 /*
diff --git a/drivers/gpu/nvgpu/os/linux/fecs_trace_linux.c b/drivers/gpu/nvgpu/os/linux/fecs_trace_linux.c
index ad83c540e..65114f272 100644
--- a/drivers/gpu/nvgpu/os/linux/fecs_trace_linux.c
+++ b/drivers/gpu/nvgpu/os/linux/fecs_trace_linux.c
@@ -508,22 +508,42 @@ static struct vm_operations_struct gk20a_ctxsw_dev_vma_ops = {
 	.close = gk20a_ctxsw_dev_vma_close,
 };
 
-int nvgpu_gr_fecs_trace_mmap_buffer(struct gk20a *g,
-				struct vm_area_struct *vma)
+void nvgpu_gr_fecs_trace_get_mmap_buffer_info(struct gk20a *g,
+				void **mmapaddr, size_t *mmapsize)
 {
-	return remap_vmalloc_range(vma, g->ctxsw_trace->devs[0].hdr, 0);
+	*mmapaddr = g->ctxsw_trace->devs[0].hdr;
+	*mmapsize = 0;
 }
 
 int gk20a_ctxsw_dev_mmap(struct file *filp, struct vm_area_struct *vma)
 {
 	struct gk20a_ctxsw_dev *dev = filp->private_data;
 	struct gk20a *g = dev->g;
+	size_t mmapsize = 0;
+	void *mmapaddr;
 	int ret;
 
 	nvgpu_log(g, gpu_dbg_fn|gpu_dbg_ctxsw, "vm_start=%lx vm_end=%lx",
 		vma->vm_start, vma->vm_end);
 
-	ret = dev->g->ops.gr.fecs_trace.mmap_user_buffer(dev->g, vma);
+	dev->g->ops.gr.fecs_trace.get_mmap_user_buffer_info(dev->g,
+					&mmapaddr, &mmapsize);
+	if (mmapsize) {
+		unsigned long size = 0;
+		unsigned long vsize = vma->vm_end - vma->vm_start;
+
+		size = min(mmapsize, vsize);
+		size = round_up(size, PAGE_SIZE);
+
+		ret = remap_pfn_range(vma, vma->vm_start,
+			(unsigned long) mmapaddr,
+			size,
+			vma->vm_page_prot);
+
+	} else {
+		ret = remap_vmalloc_range(vma, mmapaddr, 0);
+	}
+
 	if (likely(!ret)) {
 		vma->vm_private_data = dev;
 		vma->vm_ops = &gk20a_ctxsw_dev_vma_ops;
@@ -757,7 +777,7 @@ void nvgpu_gr_fecs_trace_add_tsg_reset(struct gk20a *g, struct tsg_gk20a *tsg)
 
 u8 nvgpu_gpu_ctxsw_tags_to_common_tags(u8 tags)
 {
-	switch (tags){
+	switch (tags) {
 	case NVGPU_CTXSW_TAG_SOF:
 		return NVGPU_GPU_CTXSW_TAG_SOF;
 	case NVGPU_CTXSW_TAG_CTXSW_REQ_BY_HOST:
diff --git a/drivers/gpu/nvgpu/os/linux/vgpu/fecs_trace_vgpu_linux.c b/drivers/gpu/nvgpu/os/linux/vgpu/fecs_trace_vgpu_linux.c
index 09c89df6f..e893a40ee 100644
--- a/drivers/gpu/nvgpu/os/linux/vgpu/fecs_trace_vgpu_linux.c
+++ b/drivers/gpu/nvgpu/os/linux/vgpu/fecs_trace_vgpu_linux.c
@@ -37,17 +37,11 @@ int vgpu_alloc_user_buffer(struct gk20a *g, void **buf, size_t *size)
 	return 0;
 }
 
-int vgpu_mmap_user_buffer(struct gk20a *g, struct vm_area_struct *vma)
+void vgpu_get_mmap_user_buffer_info(struct gk20a *g,
+				void **mmapaddr, size_t *mmapsize)
 {
 	struct vgpu_fecs_trace *vcst = (struct vgpu_fecs_trace *)g->fecs_trace;
-	unsigned long size = vgpu_ivm_get_size(vcst->cookie);
-	unsigned long vsize = vma->vm_end - vma->vm_start;
 
-	size = min(size, vsize);
-	size = round_up(size, PAGE_SIZE);
-
-	return remap_pfn_range(vma, vma->vm_start,
-			vgpu_ivm_get_ipa(vcst->cookie) >> PAGE_SHIFT,
-			size,
-			vma->vm_page_prot);
+	*mmapsize = vgpu_ivm_get_size(vcst->cookie);
+	*mmapaddr = (void *) (vgpu_ivm_get_ipa(vcst->cookie) >> PAGE_SHIFT);
 }
diff --git a/drivers/gpu/nvgpu/tu104/hal_tu104.c b/drivers/gpu/nvgpu/tu104/hal_tu104.c
index 73a8e4a29..de0c3ea64 100644
--- a/drivers/gpu/nvgpu/tu104/hal_tu104.c
+++ b/drivers/gpu/nvgpu/tu104/hal_tu104.c
@@ -590,7 +590,8 @@ static const struct gpu_ops tu104_ops = {
 		.fecs_trace = {
 			.alloc_user_buffer = nvgpu_gr_fecs_trace_ring_alloc,
 			.free_user_buffer = nvgpu_gr_fecs_trace_ring_free,
-			.mmap_user_buffer = nvgpu_gr_fecs_trace_mmap_buffer,
+			.get_mmap_user_buffer_info =
+				nvgpu_gr_fecs_trace_get_mmap_buffer_info,
 			.init = nvgpu_gr_fecs_trace_init,
 			.deinit = nvgpu_gr_fecs_trace_deinit,
 			.enable = nvgpu_gr_fecs_trace_enable,