nv-tegra-mirror/linux-nvgpu
1
0
Fork 0
mirror of git://nv-tegra.nvidia.com/linux-nvgpu.git synced 2025-12-22 09:12:24 +03:00
Code Issues Packages Projects Releases Wiki Activity
Files
56f56b5cd9d2e75cf7d2613b5e115bfebdbee0ce
linux-nvgpu/drivers/gpu/nvgpu/gp10b/hal_gp10b.c
Thomas Fleury 56f56b5cd9 gpu: nvgpu: hal for timestamps correlation 
In order to perform timestamps correlation for FECS
traces, we need to collect GPU / GPU timestamps
samples. In virtualization case, it is possible for
a guest to get GPU timestamps by using read_ptimer.
However, if the CPU timestamp is read on guest side,
and the GPU timestamp is read on vm-server side,
then it introduces some latency that will create an
artificial offset for GPU timestamps (~2 us in
average). For better CPU / GPU timestamps correlation,
Added a command to collect all timestamps on vm-server
side.

Bug 1900475

Change-Id: Idfdc6ae4c16c501dc5e00053a5b75932c55148d6
Signed-off-by: Thomas Fleury <tfleury@nvidia.com>
Reviewed-on: http://git-master/r/1472447
Reviewed-by: svccoveritychecker <svccoveritychecker@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Aparna Das <aparnad@nvidia.com>
Reviewed-by: Terje Bergstrom <tbergstrom@nvidia.com>
2017-05-23 14:14:46 -07:00

272 lines
7.3 KiB
C
Raw Blame History

/*
* GP10B Tegra HAL interface
*
* Copyright (c) 2014-2017, NVIDIA CORPORATION. All rights reserved.
*
* This program is free software; you can redistribute it and/or modify it
* under the terms and conditions of the GNU General Public License,
* version 2, as published by the Free Software Foundation.
*
* This program is distributed in the hope it will be useful, but WITHOUT
* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
* FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
* more details.
*/
#include <linux/types.h>
#include <linux/printk.h>
#include <linux/version.h>
#include <linux/types.h>
#include <soc/tegra/fuse.h>
#include "gk20a/gk20a.h"
#include "gk20a/dbg_gpu_gk20a.h"
#include "gk20a/css_gr_gk20a.h"
#include "gk20a/ctrl_gk20a.h"
#include "gp10b/gr_gp10b.h"
#include "gp10b/fecs_trace_gp10b.h"
#include "gp10b/mc_gp10b.h"
#include "gp10b/ltc_gp10b.h"
#include "gp10b/mm_gp10b.h"
#include "gp10b/ce_gp10b.h"
#include "gp10b/fb_gp10b.h"
#include "gp10b/pmu_gp10b.h"
#include "gp10b/gr_ctx_gp10b.h"
#include "gp10b/fifo_gp10b.h"
#include "gp10b/gp10b_gating_reglist.h"
#include "gp10b/regops_gp10b.h"
#include "gp10b/cde_gp10b.h"
#include "gp10b/therm_gp10b.h"
#include "gm20b/gr_gm20b.h"
#include "gm20b/fifo_gm20b.h"
#include "gm20b/pmu_gm20b.h"
#include "gm20b/clk_gm20b.h"
#include "gp10b.h"
#include <nvgpu/hw/gp10b/hw_proj_gp10b.h>
#if LINUX_VERSION_CODE < KERNEL_VERSION(4, 4, 0)
#define FUSE_OPT_PRIV_SEC_EN_0 0x264
#endif
#define PRIV_SECURITY_ENABLED 0x01
static struct gpu_ops gp10b_ops = {
.clock_gating = {
.slcg_bus_load_gating_prod =
gp10b_slcg_bus_load_gating_prod,
.slcg_ce2_load_gating_prod =
gp10b_slcg_ce2_load_gating_prod,
.slcg_chiplet_load_gating_prod =
gp10b_slcg_chiplet_load_gating_prod,
.slcg_ctxsw_firmware_load_gating_prod =
gp10b_slcg_ctxsw_firmware_load_gating_prod,
.slcg_fb_load_gating_prod =
gp10b_slcg_fb_load_gating_prod,
.slcg_fifo_load_gating_prod =
gp10b_slcg_fifo_load_gating_prod,
.slcg_gr_load_gating_prod =
gr_gp10b_slcg_gr_load_gating_prod,
.slcg_ltc_load_gating_prod =
ltc_gp10b_slcg_ltc_load_gating_prod,
.slcg_perf_load_gating_prod =
gp10b_slcg_perf_load_gating_prod,
.slcg_priring_load_gating_prod =
gp10b_slcg_priring_load_gating_prod,
.slcg_pmu_load_gating_prod =
gp10b_slcg_pmu_load_gating_prod,
.slcg_therm_load_gating_prod =
gp10b_slcg_therm_load_gating_prod,
.slcg_xbar_load_gating_prod =
gp10b_slcg_xbar_load_gating_prod,
.blcg_bus_load_gating_prod =
gp10b_blcg_bus_load_gating_prod,
.blcg_ce_load_gating_prod =
gp10b_blcg_ce_load_gating_prod,
.blcg_ctxsw_firmware_load_gating_prod =
gp10b_blcg_ctxsw_firmware_load_gating_prod,
.blcg_fb_load_gating_prod =
gp10b_blcg_fb_load_gating_prod,
.blcg_fifo_load_gating_prod =
gp10b_blcg_fifo_load_gating_prod,
.blcg_gr_load_gating_prod =
gp10b_blcg_gr_load_gating_prod,
.blcg_ltc_load_gating_prod =
gp10b_blcg_ltc_load_gating_prod,
.blcg_pwr_csb_load_gating_prod =
gp10b_blcg_pwr_csb_load_gating_prod,
.blcg_pmu_load_gating_prod =
gp10b_blcg_pmu_load_gating_prod,
.blcg_xbar_load_gating_prod =
gp10b_blcg_xbar_load_gating_prod,
.pg_gr_load_gating_prod =
gr_gp10b_pg_gr_load_gating_prod,
}
};
static int gp10b_get_litter_value(struct gk20a *g, int value)
{
int ret = EINVAL;
switch (value) {
case GPU_LIT_NUM_GPCS:
ret = proj_scal_litter_num_gpcs_v();
break;
case GPU_LIT_NUM_PES_PER_GPC:
ret = proj_scal_litter_num_pes_per_gpc_v();
break;
case GPU_LIT_NUM_ZCULL_BANKS:
ret = proj_scal_litter_num_zcull_banks_v();
break;
case GPU_LIT_NUM_TPC_PER_GPC:
ret = proj_scal_litter_num_tpc_per_gpc_v();
break;
case GPU_LIT_NUM_FBPS:
ret = proj_scal_litter_num_fbps_v();
break;
case GPU_LIT_GPC_BASE:
ret = proj_gpc_base_v();
break;
case GPU_LIT_GPC_STRIDE:
ret = proj_gpc_stride_v();
break;
case GPU_LIT_GPC_SHARED_BASE:
ret = proj_gpc_shared_base_v();
break;
case GPU_LIT_TPC_IN_GPC_BASE:
ret = proj_tpc_in_gpc_base_v();
break;
case GPU_LIT_TPC_IN_GPC_STRIDE:
ret = proj_tpc_in_gpc_stride_v();
break;
case GPU_LIT_TPC_IN_GPC_SHARED_BASE:
ret = proj_tpc_in_gpc_shared_base_v();
break;
case GPU_LIT_PPC_IN_GPC_BASE:
ret = proj_ppc_in_gpc_base_v();
break;
case GPU_LIT_PPC_IN_GPC_STRIDE:
ret = proj_ppc_in_gpc_stride_v();
break;
case GPU_LIT_PPC_IN_GPC_SHARED_BASE:
ret = proj_ppc_in_gpc_shared_base_v();
break;
case GPU_LIT_ROP_BASE:
ret = proj_rop_base_v();
break;
case GPU_LIT_ROP_STRIDE:
ret = proj_rop_stride_v();
break;
case GPU_LIT_ROP_SHARED_BASE:
ret = proj_rop_shared_base_v();
break;
case GPU_LIT_HOST_NUM_ENGINES:
ret = proj_host_num_engines_v();
break;
case GPU_LIT_HOST_NUM_PBDMA:
ret = proj_host_num_pbdma_v();
break;
case GPU_LIT_LTC_STRIDE:
ret = proj_ltc_stride_v();
break;
case GPU_LIT_LTS_STRIDE:
ret = proj_lts_stride_v();
break;
/* GP10B does not have a FBPA unit, despite what's listed in the
* hw headers or read back through NV_PTOP_SCAL_NUM_FBPAS,
* so hardcode all values to 0.
*/
case GPU_LIT_NUM_FBPAS:
case GPU_LIT_FBPA_STRIDE:
case GPU_LIT_FBPA_BASE:
case GPU_LIT_FBPA_SHARED_BASE:
ret = 0;
break;
default:
gk20a_err(dev_from_gk20a(g), "Missing definition %d", value);
BUG();
break;
}
return ret;
}
int gp10b_init_hal(struct gk20a *g)
{
struct gpu_ops *gops = &g->ops;
struct nvgpu_gpu_characteristics *c = &g->gpu_characteristics;
struct gk20a_platform *platform = dev_get_drvdata(g->dev);
u32 val;
*gops = gp10b_ops;
gops->pmupstate = false;
#ifdef CONFIG_TEGRA_ACR
if (platform->is_fmodel) {
gops->privsecurity = 0;
gops->securegpccs = 0;
} else {
tegra_fuse_readl(FUSE_OPT_PRIV_SEC_EN_0, &val);
if (val & PRIV_SECURITY_ENABLED) {
gops->privsecurity = 1;
gops->securegpccs =1;
} else {
gk20a_dbg_info("priv security is disabled in HW");
gops->privsecurity = 0;
gops->securegpccs = 0;
}
}
#else
if (platform->is_fmodel) {
gk20a_dbg_info("running simulator with PRIV security disabled");
gops->privsecurity = 0;
gops->securegpccs = 0;
} else {
tegra_fuse_readl(FUSE_OPT_PRIV_SEC_EN_0, &val);
if (val & PRIV_SECURITY_ENABLED) {
gk20a_dbg_info("priv security is not supported but enabled");
gops->privsecurity = 1;
gops->securegpccs =1;
return -EPERM;
} else {
gops->privsecurity = 0;
gops->securegpccs = 0;
}
}
#endif
gp10b_init_mc(gops);
gp10b_init_gr(gops);
gp10b_init_fecs_trace_ops(gops);
gp10b_init_ltc(gops);
gp10b_init_fb(gops);
gp10b_init_fifo(gops);
gp10b_init_ce(gops);
gp10b_init_gr_ctx(gops);
gp10b_init_mm(gops);
gp10b_init_pmu_ops(gops);
gk20a_init_debug_ops(gops);
gk20a_init_dbg_session_ops(gops);
gp10b_init_regops(gops);
gp10b_init_cde_ops(gops);
gp10b_init_therm_ops(gops);
gk20a_init_tsg_ops(gops);
#if defined(CONFIG_GK20A_CYCLE_STATS)
gk20a_init_css_ops(gops);
#endif
gops->name = "gp10b";
gops->chip_init_gpu_characteristics = gp10b_init_gpu_characteristics;
gops->get_litter_value = gp10b_get_litter_value;
gops->read_ptimer = gk20a_read_ptimer;
gops->bus.get_timestamps_zipper = gk20a_get_timestamps_zipper;
c->twod_class = FERMI_TWOD_A;
c->threed_class = PASCAL_A;
c->compute_class = PASCAL_COMPUTE_A;
c->gpfifo_class = PASCAL_CHANNEL_GPFIFO_A;
c->inline_to_memory_class = KEPLER_INLINE_TO_MEMORY_B;
c->dma_copy_class = PASCAL_DMA_COPY_A;
return 0;
}
Reference in New Issue View Git Blame Copy Permalink