cachepc-linux

Fork of AMDESE/linux with modifications for CachePC side-channel attack
git clone https://git.sinitax.com/sinitax/cachepc-linux
Log | Files | Refs | README | LICENSE | sfeed.txt

gk20a.c (5798B)


      1/*
      2 * Copyright (c) 2014-2016, NVIDIA CORPORATION. All rights reserved.
      3 *
      4 * Permission is hereby granted, free of charge, to any person obtaining a
      5 * copy of this software and associated documentation files (the "Software"),
      6 * to deal in the Software without restriction, including without limitation
      7 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
      8 * and/or sell copies of the Software, and to permit persons to whom the
      9 * Software is furnished to do so, subject to the following conditions:
     10 *
     11 * The above copyright notice and this permission notice shall be included in
     12 * all copies or substantial portions of the Software.
     13 *
     14 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
     15 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
     16 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
     17 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
     18 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
     19 * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
     20 * DEALINGS IN THE SOFTWARE.
     21 */
     22#define gk20a_volt(p) container_of((p), struct gk20a_volt, base)
     23#include "priv.h"
     24
     25#include <core/tegra.h>
     26
     27#include "gk20a.h"
     28
     29static const struct cvb_coef gk20a_cvb_coef[] = {
     30	/* MHz,        c0,     c1,   c2,    c3,     c4,   c5 */
     31	/*  72 */ { 1209886, -36468,  515,   417, -13123,  203},
     32	/* 108 */ { 1130804, -27659,  296,   298, -10834,  221},
     33	/* 180 */ { 1162871, -27110,  247,   238, -10681,  268},
     34	/* 252 */ { 1220458, -28654,  247,   179, -10376,  298},
     35	/* 324 */ { 1280953, -30204,  247,   119,  -9766,  304},
     36	/* 396 */ { 1344547, -31777,  247,   119,  -8545,  292},
     37	/* 468 */ { 1420168, -34227,  269,    60,  -7172,  256},
     38	/* 540 */ { 1490757, -35955,  274,    60,  -5188,  197},
     39	/* 612 */ { 1599112, -42583,  398,     0,  -1831,  119},
     40	/* 648 */ { 1366986, -16459, -274,     0,  -3204,   72},
     41	/* 684 */ { 1391884, -17078, -274,   -60,  -1526,   30},
     42	/* 708 */ { 1415522, -17497, -274,   -60,   -458,    0},
     43	/* 756 */ { 1464061, -18331, -274,  -119,   1831,  -72},
     44	/* 804 */ { 1524225, -20064, -254,  -119,   4272, -155},
     45	/* 852 */ { 1608418, -21643, -269,     0,    763,  -48},
     46};
     47
     48/**
     49 * cvb_mv = ((c2 * speedo / s_scale + c1) * speedo / s_scale + c0)
     50 */
     51static inline int
     52gk20a_volt_get_cvb_voltage(int speedo, int s_scale, const struct cvb_coef *coef)
     53{
     54	int mv;
     55
     56	mv = DIV_ROUND_CLOSEST(coef->c2 * speedo, s_scale);
     57	mv = DIV_ROUND_CLOSEST((mv + coef->c1) * speedo, s_scale) + coef->c0;
     58	return mv;
     59}
     60
     61/**
     62 * cvb_t_mv =
     63 * ((c2 * speedo / s_scale + c1) * speedo / s_scale + c0) +
     64 * ((c3 * speedo / s_scale + c4 + c5 * T / t_scale) * T / t_scale)
     65 */
     66static inline int
     67gk20a_volt_get_cvb_t_voltage(int speedo, int temp, int s_scale, int t_scale,
     68			     const struct cvb_coef *coef)
     69{
     70	int cvb_mv, mv;
     71
     72	cvb_mv = gk20a_volt_get_cvb_voltage(speedo, s_scale, coef);
     73
     74	mv = DIV_ROUND_CLOSEST(coef->c3 * speedo, s_scale) + coef->c4 +
     75		DIV_ROUND_CLOSEST(coef->c5 * temp, t_scale);
     76	mv = DIV_ROUND_CLOSEST(mv * temp, t_scale) + cvb_mv;
     77	return mv;
     78}
     79
     80static int
     81gk20a_volt_calc_voltage(const struct cvb_coef *coef, int speedo)
     82{
     83	static const int v_scale = 1000;
     84	int mv;
     85
     86	mv = gk20a_volt_get_cvb_t_voltage(speedo, -10, 100, 10, coef);
     87	mv = DIV_ROUND_UP(mv, v_scale);
     88
     89	return mv * 1000;
     90}
     91
     92static int
     93gk20a_volt_vid_get(struct nvkm_volt *base)
     94{
     95	struct gk20a_volt *volt = gk20a_volt(base);
     96	int i, uv;
     97
     98	uv = regulator_get_voltage(volt->vdd);
     99
    100	for (i = 0; i < volt->base.vid_nr; i++)
    101		if (volt->base.vid[i].uv >= uv)
    102			return i;
    103
    104	return -EINVAL;
    105}
    106
    107static int
    108gk20a_volt_vid_set(struct nvkm_volt *base, u8 vid)
    109{
    110	struct gk20a_volt *volt = gk20a_volt(base);
    111	struct nvkm_subdev *subdev = &volt->base.subdev;
    112
    113	nvkm_debug(subdev, "set voltage as %duv\n", volt->base.vid[vid].uv);
    114	return regulator_set_voltage(volt->vdd, volt->base.vid[vid].uv, 1200000);
    115}
    116
    117static int
    118gk20a_volt_set_id(struct nvkm_volt *base, u8 id, int condition)
    119{
    120	struct gk20a_volt *volt = gk20a_volt(base);
    121	struct nvkm_subdev *subdev = &volt->base.subdev;
    122	int prev_uv = regulator_get_voltage(volt->vdd);
    123	int target_uv = volt->base.vid[id].uv;
    124	int ret;
    125
    126	nvkm_debug(subdev, "prev=%d, target=%d, condition=%d\n",
    127		   prev_uv, target_uv, condition);
    128	if (!condition ||
    129		(condition < 0 && target_uv < prev_uv) ||
    130		(condition > 0 && target_uv > prev_uv)) {
    131		ret = gk20a_volt_vid_set(&volt->base, volt->base.vid[id].vid);
    132	} else {
    133		ret = 0;
    134	}
    135
    136	return ret;
    137}
    138
    139static const struct nvkm_volt_func
    140gk20a_volt = {
    141	.vid_get = gk20a_volt_vid_get,
    142	.vid_set = gk20a_volt_vid_set,
    143	.set_id = gk20a_volt_set_id,
    144};
    145
    146int
    147gk20a_volt_ctor(struct nvkm_device *device, enum nvkm_subdev_type type, int inst,
    148		const struct cvb_coef *coefs, int nb_coefs,
    149		int vmin, struct gk20a_volt *volt)
    150{
    151	struct nvkm_device_tegra *tdev = device->func->tegra(device);
    152	int i, uv;
    153
    154	nvkm_volt_ctor(&gk20a_volt, device, type, inst, &volt->base);
    155
    156	uv = regulator_get_voltage(tdev->vdd);
    157	nvkm_debug(&volt->base.subdev, "the default voltage is %duV\n", uv);
    158
    159	volt->vdd = tdev->vdd;
    160
    161	volt->base.vid_nr = nb_coefs;
    162	for (i = 0; i < volt->base.vid_nr; i++) {
    163		volt->base.vid[i].vid = i;
    164		volt->base.vid[i].uv = max(
    165			gk20a_volt_calc_voltage(&coefs[i], tdev->gpu_speedo),
    166			vmin);
    167		nvkm_debug(&volt->base.subdev, "%2d: vid=%d, uv=%d\n", i,
    168			   volt->base.vid[i].vid, volt->base.vid[i].uv);
    169	}
    170
    171	return 0;
    172}
    173
    174int
    175gk20a_volt_new(struct nvkm_device *device, enum nvkm_subdev_type type, int inst, struct nvkm_volt **pvolt)
    176{
    177	struct gk20a_volt *volt;
    178
    179	volt = kzalloc(sizeof(*volt), GFP_KERNEL);
    180	if (!volt)
    181		return -ENOMEM;
    182	*pvolt = &volt->base;
    183
    184	return gk20a_volt_ctor(device, type, inst, gk20a_cvb_coef,
    185			       ARRAY_SIZE(gk20a_cvb_coef), 0, volt);
    186}