Files
linux-nvgpu/drivers/gpu/nvgpu/gk20a/channel_sync_gk20a.h
Deepak Nibade 0c46f8a5e1 gpu: nvgpu: support user fence updates
Add support for user fence updates i.e. increments added by user space
in pushbuffer directly

Add a submit IOCTL flag NVGPU_SUBMIT_GPFIFO_FLAGS_USER_FENCE_UPDATE to indicate
if User has added increments in pushbuffer
If yes, number_of_increment value is received in fence.value from User

If User is adding increments in the pushbuffer then we don't need to do any job
tracking in the kernel
So fail the submit if we evaluate need_job_tracking to true and
FLAGS_USER_FENCE_UPDATE is set
User is responsible for ensuring all pre-requisites for a fast submit and to
prevent kernel job tracking

Since user space adds increments in the pushbuffer, just handle the threshold
book keeping in kernel.

Bug 200326065
Jira NVGPU-179

Change-Id: Ic0f0b1aa69e3389a4c3305fb6a559c5113719e0f
Signed-off-by: Deepak Nibade <dnibade@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1661854
Reviewed-by: svc-mobile-coverity <svc-mobile-coverity@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Vijayakumar Subbu <vsubbu@nvidia.com>
Reviewed-by: Konsta Holtta <kholtta@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
2018-02-26 03:48:14 -08:00

128 lines
4.5 KiB
C

/*
* drivers/video/tegra/host/gk20a/channel_sync_gk20a.h
*
* GK20A Channel Synchronization Abstraction
*
* Copyright (c) 2014-2017, 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"),
* to deal in the Software without restriction, including without limitation
* the rights to use, copy, modify, merge, publish, distribute, sublicense,
* and/or sell copies of the Software, and to permit persons to whom the
* Software is furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in
* all copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
* THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
* FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
* DEALINGS IN THE SOFTWARE.
*/
#ifndef _GK20A_CHANNEL_SYNC_H_
#define _GK20A_CHANNEL_SYNC_H_
struct gk20a_channel_sync;
struct priv_cmd_entry;
struct channel_gk20a;
struct gk20a_fence;
struct gk20a;
struct gk20a_channel_sync {
nvgpu_atomic_t refcount;
/* Generate a gpu wait cmdbuf from syncpoint.
* Returns
* - a gpu cmdbuf that performs the wait when executed,
* - possibly a helper fence that the caller must hold until the
* cmdbuf is executed.
*/
int (*wait_syncpt)(struct gk20a_channel_sync *s, u32 id, u32 thresh,
struct priv_cmd_entry *entry,
struct gk20a_fence *fence);
/* Generate a gpu wait cmdbuf from sync fd.
* Returns
* - a gpu cmdbuf that performs the wait when executed,
* - possibly a helper fence that the caller must hold until the
* cmdbuf is executed.
*/
int (*wait_fd)(struct gk20a_channel_sync *s, int fd,
struct priv_cmd_entry *entry,
struct gk20a_fence *fence);
/* Increment syncpoint/semaphore.
* Returns
* - a gpu cmdbuf that performs the increment when executed,
* - a fence that can be passed to wait_cpu() and is_expired().
*/
int (*incr)(struct gk20a_channel_sync *s,
struct priv_cmd_entry *entry,
struct gk20a_fence *fence,
bool need_sync_fence,
bool register_irq);
/* Increment syncpoint/semaphore, preceded by a wfi.
* Returns
* - a gpu cmdbuf that performs the increment when executed,
* - a fence that can be passed to wait_cpu() and is_expired().
*/
int (*incr_wfi)(struct gk20a_channel_sync *s,
struct priv_cmd_entry *entry,
struct gk20a_fence *fence);
/* Increment syncpoint/semaphore, so that the returned fence represents
* work completion (may need wfi) and can be returned to user space.
* Returns
* - a gpu cmdbuf that performs the increment when executed,
* - a fence that can be passed to wait_cpu() and is_expired(),
* - a gk20a_fence that signals when the incr has happened.
*/
int (*incr_user)(struct gk20a_channel_sync *s,
int wait_fence_fd,
struct priv_cmd_entry *entry,
struct gk20a_fence *fence,
bool wfi,
bool need_sync_fence,
bool register_irq);
/* Reset the channel syncpoint/semaphore. */
void (*set_min_eq_max)(struct gk20a_channel_sync *s);
/* Signals the sync timeline (if owned by the gk20a_channel_sync layer).
* This should be called when we notice that a gk20a_fence is
* expired. */
void (*signal_timeline)(struct gk20a_channel_sync *s);
/* Returns the sync point id or negative number if no syncpt*/
int (*syncpt_id)(struct gk20a_channel_sync *s);
/* 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);
};
void gk20a_channel_sync_destroy(struct gk20a_channel_sync *sync);
struct gk20a_channel_sync *gk20a_channel_sync_create(struct channel_gk20a *c);
bool gk20a_channel_sync_needs_sync_framework(struct gk20a *g);
#ifdef CONFIG_SYNC
void gk20a_channel_cancel_pending_sema_waits(struct gk20a *g);
#else
static inline void gk20a_channel_cancel_pending_sema_waits(struct gk20a *g)
{
}
#endif
#endif