diff --git a/drivers/gpu/nvgpu/common/linux/channel.c b/drivers/gpu/nvgpu/common/linux/channel.c
index 35fb3023d..a725cd6b0 100644
--- a/drivers/gpu/nvgpu/common/linux/channel.c
+++ b/drivers/gpu/nvgpu/common/linux/channel.c
@@ -720,8 +720,7 @@ int gk20a_submit_channel_gpfifo(struct channel_gk20a *c,
 		return -EINVAL;
 
 	if ((flags & (NVGPU_SUBMIT_GPFIFO_FLAGS_FENCE_WAIT |
-		      NVGPU_SUBMIT_GPFIFO_FLAGS_FENCE_GET |
-		      NVGPU_SUBMIT_GPFIFO_FLAGS_USER_FENCE_UPDATE)) &&
+		      NVGPU_SUBMIT_GPFIFO_FLAGS_FENCE_GET)) &&
 	    !fence)
 		return -EINVAL;
 
@@ -758,16 +757,6 @@ int gk20a_submit_channel_gpfifo(struct channel_gk20a *c,
 			(g->can_railgate && !c->deterministic) ||
 			!skip_buffer_refcounting;
 
-	/*
-	 * If User is adding increments to the pushbuffer and doing all job
-	 * tracking, then no need for kernel tracking here
-	 * User should ensure that all pre-requisites for fast submit are met
-	 * Fail the submit if that's not the case
-	 */
-	if (need_job_tracking &&
-	    (flags & NVGPU_SUBMIT_GPFIFO_FLAGS_USER_FENCE_UPDATE))
-		return -EINVAL;
-
 	if (need_job_tracking) {
 		bool need_sync_framework = false;
 
@@ -879,15 +868,6 @@ int gk20a_submit_channel_gpfifo(struct channel_gk20a *c,
 		goto clean_up;
 	}
 
-	if (flags & NVGPU_SUBMIT_GPFIFO_FLAGS_USER_FENCE_UPDATE) {
-		/*
-		 * User space adds increments in the pushbuffer, so just
-		 * handle the threshold book keeping in kernel by adding
-		 * number of syncpoint increments to threshold
-		 */
-		c->sync->add_user_incrs(c->sync, fence->value);
-	}
-
 	if (need_job_tracking) {
 		err = channel_gk20a_alloc_job(c, &job);
 		if (err)
diff --git a/drivers/gpu/nvgpu/gk20a/channel_sync_gk20a.c b/drivers/gpu/nvgpu/gk20a/channel_sync_gk20a.c
index 18d61faa2..e965a329b 100644
--- a/drivers/gpu/nvgpu/gk20a/channel_sync_gk20a.c
+++ b/drivers/gpu/nvgpu/gk20a/channel_sync_gk20a.c
@@ -301,13 +301,6 @@ static u64 gk20a_channel_syncpt_address(struct gk20a_channel_sync *s)
 	return sp->syncpt_buf.gpu_va;
 }
 
-static u32 gk20a_channel_add_user_incrs(struct gk20a_channel_sync *s, u32 val)
-{
-	struct gk20a_channel_syncpt *sp =
-		container_of(s, struct gk20a_channel_syncpt, ops);
-	return nvgpu_nvhost_syncpt_incr_max_ext(sp->nvhost_dev, sp->id, val);
-}
-
 static void gk20a_channel_syncpt_destroy(struct gk20a_channel_sync *s)
 {
 	struct gk20a_channel_syncpt *sp =
@@ -360,7 +353,6 @@ gk20a_channel_syncpt_create(struct channel_gk20a *c)
 	sp->ops.signal_timeline		= gk20a_channel_syncpt_signal_timeline;
 	sp->ops.syncpt_id		= gk20a_channel_syncpt_id;
 	sp->ops.syncpt_address		= gk20a_channel_syncpt_address;
-	sp->ops.add_user_incrs		= gk20a_channel_add_user_incrs;
 	sp->ops.destroy			= gk20a_channel_syncpt_destroy;
 
 	return &sp->ops;
@@ -886,12 +878,6 @@ static u64 gk20a_channel_semaphore_syncpt_address(struct gk20a_channel_sync *s)
 	return 0;
 }
 
-static u32 gk20a_channel_semaphore_add_user_incrs(struct gk20a_channel_sync *s,
-						  u32 val)
-{
-	return 0;
-}
-
 static void gk20a_channel_semaphore_destroy(struct gk20a_channel_sync *s)
 {
 	struct gk20a_channel_semaphore *sema =
@@ -944,7 +930,6 @@ gk20a_channel_semaphore_create(struct channel_gk20a *c)
 	sema->ops.signal_timeline = gk20a_channel_semaphore_signal_timeline;
 	sema->ops.syncpt_id	= gk20a_channel_semaphore_syncpt_id;
 	sema->ops.syncpt_address = gk20a_channel_semaphore_syncpt_address;
-	sema->ops.add_user_incrs = gk20a_channel_semaphore_add_user_incrs;
 	sema->ops.destroy	= gk20a_channel_semaphore_destroy;
 
 	return &sema->ops;
diff --git a/drivers/gpu/nvgpu/gk20a/channel_sync_gk20a.h b/drivers/gpu/nvgpu/gk20a/channel_sync_gk20a.h
index c80ebd380..fe1d8526e 100644
--- a/drivers/gpu/nvgpu/gk20a/channel_sync_gk20a.h
+++ b/drivers/gpu/nvgpu/gk20a/channel_sync_gk20a.h
@@ -105,9 +105,6 @@ struct gk20a_channel_sync {
 	/* Returns the sync point address of sync point or 0 if not supported */
 	u64 (*syncpt_address)(struct gk20a_channel_sync *s);
 
-	/* Handle user added increments in the push buffer */
-	u32 (*add_user_incrs)(struct gk20a_channel_sync *s, u32 val);
-
 	/* Free the resources allocated by gk20a_channel_sync_create. */
 	void (*destroy)(struct gk20a_channel_sync *s);
 };
diff --git a/include/uapi/linux/nvgpu.h b/include/uapi/linux/nvgpu.h
index 18168158b..cf75595a8 100644
--- a/include/uapi/linux/nvgpu.h
+++ b/include/uapi/linux/nvgpu.h
@@ -1478,8 +1478,6 @@ struct nvgpu_fence {
 #define NVGPU_SUBMIT_GPFIFO_FLAGS_SKIP_BUFFER_REFCOUNTING	(1 << 5)
 /* expire current timeslice and reschedule runlist from front */
 #define NVGPU_SUBMIT_GPFIFO_FLAGS_RESCHEDULE_RUNLIST	(1 << 6)
-/* user space has added syncpoint increments in the pushbuffer */
-#define NVGPU_SUBMIT_GPFIFO_FLAGS_USER_FENCE_UPDATE	(1 << 7)
 
 struct nvgpu_submit_gpfifo_args {
 	__u64 gpfifo;