diff --git a/drivers/gpu/nvgpu/common/mm/gmmu/page_table.c b/drivers/gpu/nvgpu/common/mm/gmmu/page_table.c
index 42393b688..da35abb26 100644
--- a/drivers/gpu/nvgpu/common/mm/gmmu/page_table.c
+++ b/drivers/gpu/nvgpu/common/mm/gmmu/page_table.c
@@ -713,7 +713,7 @@ static int __nvgpu_gmmu_update_page_table(struct vm_gk20a *vm,
 		   "vm=%s "
 		   "%-5s GPU virt %#-12llx +%#-9llx    phys %#-12llx "
 		   "phys offset: %#-4llx;  pgsz: %3dkb perm=%-2s | "
-		   "kind=%#02x APT=%-6s %c%c%c%c%c",
+		   "kind=%#02x APT=%-6s %c%c%c%c",
 		   vm->name,
 		   (sgt != NULL) ? "MAP" : "UNMAP",
 		   virt_addr,
@@ -727,7 +727,6 @@ static int __nvgpu_gmmu_update_page_table(struct vm_gk20a *vm,
 		   attrs->cacheable ? 'C' : '-',
 		   attrs->sparse    ? 'S' : '-',
 		   attrs->priv      ? 'P' : '-',
-		   attrs->coherent  ? 'I' : '-',
 		   attrs->valid     ? 'V' : '-');
 
 	err = __nvgpu_gmmu_do_update_page_table(vm,
@@ -785,7 +784,6 @@ u64 gk20a_locked_gmmu_map(struct vm_gk20a *vm,
 		.rw_flag   = rw_flag,
 		.sparse    = sparse,
 		.priv      = priv,
-		.coherent  = flags & NVGPU_VM_MAP_IO_COHERENT,
 		.valid     = (flags & NVGPU_VM_MAP_UNMAPPED_PTE) == 0U,
 		.aperture  = aperture
 	};
@@ -801,14 +799,6 @@ u64 gk20a_locked_gmmu_map(struct vm_gk20a *vm,
 
 	attrs.l3_alloc = (bool)(flags & NVGPU_VM_MAP_L3_ALLOC);
 
-	/*
-	 * Handle the IO coherency aperture: make sure the .aperture field is
-	 * correct based on the IO coherency flag.
-	 */
-	if (attrs.coherent && attrs.aperture == APERTURE_SYSMEM) {
-		attrs.aperture = APERTURE_SYSMEM_COH;
-	}
-
 	/*
 	 * Only allocate a new GPU VA range if we haven't already been passed a
 	 * GPU VA range. This facilitates fixed mappings.
@@ -866,7 +856,6 @@ void gk20a_locked_gmmu_unmap(struct vm_gk20a *vm,
 		.rw_flag   = rw_flag,
 		.sparse    = sparse,
 		.priv      = 0,
-		.coherent  = 0,
 		.valid     = 0,
 		.aperture  = APERTURE_INVALID,
 	};
diff --git a/drivers/gpu/nvgpu/gk20a/mm_gk20a.c b/drivers/gpu/nvgpu/gk20a/mm_gk20a.c
index 7ca07fc6b..d6ca478a7 100644
--- a/drivers/gpu/nvgpu/gk20a/mm_gk20a.c
+++ b/drivers/gpu/nvgpu/gk20a/mm_gk20a.c
@@ -288,7 +288,7 @@ static void update_gmmu_pte_locked(struct vm_gk20a *vm,
 	pte_dbg(g, attrs,
 		"PTE: i=%-4u size=%-2u offs=%-4u | "
 		"GPU %#-12llx  phys %#-12llx "
-		"pgsz: %3dkb perm=%-2s kind=%#02x APT=%-6s %c%c%c%c%c "
+		"pgsz: %3dkb perm=%-2s kind=%#02x APT=%-6s %c%c%c%c "
 		"ctag=0x%08x "
 		"[0x%08x, 0x%08x]",
 		pd_idx, l->entry_size, pd_offset,
@@ -300,7 +300,6 @@ static void update_gmmu_pte_locked(struct vm_gk20a *vm,
 		attrs->cacheable ? 'C' : '-',
 		attrs->sparse    ? 'S' : '-',
 		attrs->priv      ? 'P' : '-',
-		attrs->coherent  ? 'I' : '-',
 		attrs->valid     ? 'V' : '-',
 		U32(attrs->ctag) >> U32(ctag_shift),
 		pte_w[1], pte_w[0]);
diff --git a/drivers/gpu/nvgpu/gp10b/mm_gp10b.c b/drivers/gpu/nvgpu/gp10b/mm_gp10b.c
index 043d4a7f2..080e899c8 100644
--- a/drivers/gpu/nvgpu/gp10b/mm_gp10b.c
+++ b/drivers/gpu/nvgpu/gp10b/mm_gp10b.c
@@ -254,7 +254,7 @@ static void update_gmmu_pte_locked(struct vm_gk20a *vm,
 		"vm=%s "
 		"PTE: i=%-4u size=%-2u | "
 		"GPU %#-12llx  phys %#-12llx "
-		"pgsz: %3dkb perm=%-2s kind=%#02x APT=%-6s %c%c%c%c%c "
+		"pgsz: %3dkb perm=%-2s kind=%#02x APT=%-6s %c%c%c%c "
 		"ctag=0x%08x "
 		"[0x%08x, 0x%08x]",
 		vm->name,
@@ -267,7 +267,6 @@ static void update_gmmu_pte_locked(struct vm_gk20a *vm,
 		attrs->cacheable ? 'C' : '-',
 		attrs->sparse    ? 'S' : '-',
 		attrs->priv      ? 'P' : '-',
-		attrs->coherent  ? 'I' : '-',
 		attrs->valid     ? 'V' : '-',
 		(u32)attrs->ctag / g->ops.fb.compression_page_size(g),
 		pte_w[1], pte_w[0]);
diff --git a/drivers/gpu/nvgpu/include/nvgpu/gmmu.h b/drivers/gpu/nvgpu/include/nvgpu/gmmu.h
index d7db66fd4..3ad0b80da 100644
--- a/drivers/gpu/nvgpu/include/nvgpu/gmmu.h
+++ b/drivers/gpu/nvgpu/include/nvgpu/gmmu.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"),
@@ -63,7 +63,6 @@ enum gk20a_mem_rw_flag {
  *   rw_flag:     Flag from enum gk20a_mem_rw_flag
  *   sparse:      Set if the mapping should be sparse.
  *   priv:        Privilidged mapping.
- *   coherent:    Set if the mapping should be IO coherent.
  *   valid:       Set if the PTE should be marked valid.
  *   aperture:    VIDMEM or SYSMEM.
  *   debug:       When set print debugging info.
@@ -81,7 +80,6 @@ struct nvgpu_gmmu_attrs {
 	enum gk20a_mem_rw_flag	 rw_flag;
 	bool			 sparse;
 	bool			 priv;
-	bool			 coherent;
 	bool			 valid;
 	enum nvgpu_aperture	 aperture;
 	bool			 debug;
diff --git a/include/uapi/linux/nvgpu.h b/include/uapi/linux/nvgpu.h
index 51efd8f17..d6e830eb9 100644
--- a/include/uapi/linux/nvgpu.h
+++ b/include/uapi/linux/nvgpu.h
@@ -1954,6 +1954,8 @@ struct nvgpu_as_bind_channel_args {
  *
  *     Specify that a mapping shall be IO coherent.
  *
+ *     DEPRECATED: do not use! This will be removed in a future update.
+ *
  *   %NVGPU_AS_MAP_BUFFER_FLAGS_UNMAPPED_PTE
  *
  *     Specify that a mapping shall be marked as invalid but otherwise