mirror of
https://github.com/NVIDIA/open-gpu-kernel-modules.git
synced 2025-01-06 01:46:05 +01:00
252 lines
6.8 KiB
C
252 lines
6.8 KiB
C
/*
|
|
* SPDX-FileCopyrightText: Copyright (c) 2019-2021 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
|
* SPDX-License-Identifier: MIT
|
|
*
|
|
* 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 __NV_TIME_H__
|
|
#define __NV_TIME_H__
|
|
|
|
#include "conftest.h"
|
|
#include <linux/sched.h>
|
|
#include <linux/delay.h>
|
|
#include <linux/interrupt.h>
|
|
#include <linux/ktime.h>
|
|
|
|
#include <nvstatus.h>
|
|
|
|
#define NV_MAX_ISR_DELAY_US 20000
|
|
#define NV_MAX_ISR_DELAY_MS (NV_MAX_ISR_DELAY_US / 1000)
|
|
#define NV_NSECS_TO_JIFFIES(nsec) ((nsec) * HZ / 1000000000)
|
|
|
|
#if !defined(NV_TIMESPEC64_PRESENT)
|
|
struct timespec64 {
|
|
__s64 tv_sec;
|
|
long tv_nsec;
|
|
};
|
|
#endif
|
|
|
|
#if !defined(NV_KTIME_GET_RAW_TS64_PRESENT)
|
|
static inline void ktime_get_raw_ts64(struct timespec64 *ts64)
|
|
{
|
|
struct timespec ts;
|
|
getrawmonotonic(&ts);
|
|
ts64->tv_sec = ts.tv_sec;
|
|
ts64->tv_nsec = ts.tv_nsec;
|
|
}
|
|
#endif
|
|
|
|
#if !defined(NV_KTIME_GET_REAL_TS64_PRESENT)
|
|
static inline void ktime_get_real_ts64(struct timespec64 *ts64)
|
|
{
|
|
struct timeval tv;
|
|
do_gettimeofday(&tv);
|
|
ts64->tv_sec = tv.tv_sec;
|
|
ts64->tv_nsec = tv.tv_usec * (NvU64) NSEC_PER_USEC;
|
|
}
|
|
#endif
|
|
|
|
static NvBool nv_timer_less_than
|
|
(
|
|
const struct timespec64 *a,
|
|
const struct timespec64 *b
|
|
)
|
|
{
|
|
return (a->tv_sec == b->tv_sec) ? (a->tv_nsec < b->tv_nsec)
|
|
: (a->tv_sec < b->tv_sec);
|
|
}
|
|
|
|
#if !defined(NV_TIMESPEC64_PRESENT)
|
|
static inline struct timespec64 timespec64_add
|
|
(
|
|
const struct timespec64 a,
|
|
const struct timespec64 b
|
|
)
|
|
{
|
|
struct timespec64 result;
|
|
|
|
result.tv_sec = a.tv_sec + b.tv_sec;
|
|
result.tv_nsec = a.tv_nsec + b.tv_nsec;
|
|
while (result.tv_nsec >= NSEC_PER_SEC)
|
|
{
|
|
++result.tv_sec;
|
|
result.tv_nsec -= NSEC_PER_SEC;
|
|
}
|
|
return result;
|
|
}
|
|
|
|
static inline struct timespec64 timespec64_sub
|
|
(
|
|
const struct timespec64 a,
|
|
const struct timespec64 b
|
|
)
|
|
{
|
|
struct timespec64 result;
|
|
|
|
result.tv_sec = a.tv_sec - b.tv_sec;
|
|
result.tv_nsec = a.tv_nsec - b.tv_nsec;
|
|
while (result.tv_nsec < 0)
|
|
{
|
|
--(result.tv_sec);
|
|
result.tv_nsec += NSEC_PER_SEC;
|
|
}
|
|
return result;
|
|
}
|
|
|
|
static inline s64 timespec64_to_ns(struct timespec64 *ts)
|
|
{
|
|
return ((s64) ts->tv_sec * NSEC_PER_SEC) + ts->tv_nsec;
|
|
}
|
|
#endif
|
|
|
|
static inline NvU64 nv_ktime_get_raw_ns(void)
|
|
{
|
|
struct timespec64 ts;
|
|
ktime_get_raw_ts64(&ts);
|
|
return (NvU64)timespec64_to_ns(&ts);
|
|
}
|
|
|
|
// #define NV_CHECK_DELAY_ACCURACY 1
|
|
|
|
/*
|
|
* It is generally a bad idea to use udelay() to wait for more than
|
|
* a few milliseconds. Since the caller is most likely not aware of
|
|
* this, we use mdelay() for any full millisecond to be safe.
|
|
*/
|
|
static inline NV_STATUS nv_sleep_us(unsigned int us)
|
|
{
|
|
|
|
unsigned long mdelay_safe_msec;
|
|
unsigned long usec;
|
|
|
|
#ifdef NV_CHECK_DELAY_ACCURACY
|
|
struct timespec64 tm1, tm2, tm_diff;
|
|
|
|
ktime_get_raw_ts64(&tm1);
|
|
#endif
|
|
|
|
if (in_irq() && (us > NV_MAX_ISR_DELAY_US))
|
|
return NV_ERR_GENERIC;
|
|
|
|
mdelay_safe_msec = us / 1000;
|
|
if (mdelay_safe_msec)
|
|
mdelay(mdelay_safe_msec);
|
|
|
|
usec = us % 1000;
|
|
if (usec)
|
|
udelay(usec);
|
|
|
|
#ifdef NV_CHECK_DELAY_ACCURACY
|
|
ktime_get_raw_ts64(&tm2);
|
|
tm_diff = timespec64_sub(tm2, tm1);
|
|
pr_info("NVRM: delay of %d usec results in actual delay of 0x%llu nsec\n",
|
|
us, timespec64_to_ns(&tm_diff));
|
|
#endif
|
|
return NV_OK;
|
|
}
|
|
|
|
/*
|
|
* Sleep for specified milliseconds. Yields the CPU to scheduler.
|
|
*
|
|
* On Linux, a jiffie represents the time passed in between two timer
|
|
* interrupts. The number of jiffies per second (HZ) varies across the
|
|
* supported platforms. On i386, where HZ is 100, a timer interrupt is
|
|
* generated every 10ms. NV_MSECS_TO_JIFFIES should be accurate independent of
|
|
* the actual value of HZ; any partial jiffies will be 'floor'ed, the
|
|
* remainder will be accounted for with mdelay().
|
|
*/
|
|
static inline NV_STATUS nv_sleep_ms(unsigned int ms)
|
|
{
|
|
NvU64 ns;
|
|
unsigned long jiffies;
|
|
unsigned long mdelay_safe_msec;
|
|
struct timespec64 tm_end, tm_aux;
|
|
#ifdef NV_CHECK_DELAY_ACCURACY
|
|
struct timespec64 tm_start;
|
|
#endif
|
|
|
|
ktime_get_raw_ts64(&tm_aux);
|
|
#ifdef NV_CHECK_DELAY_ACCURACY
|
|
tm_start = tm_aux;
|
|
#endif
|
|
|
|
if (in_irq() && (ms > NV_MAX_ISR_DELAY_MS))
|
|
{
|
|
return NV_ERR_GENERIC;
|
|
}
|
|
|
|
if (irqs_disabled() || in_interrupt() || in_atomic())
|
|
{
|
|
mdelay(ms);
|
|
return NV_OK;
|
|
}
|
|
|
|
ns = ms * (NvU64) NSEC_PER_MSEC;
|
|
tm_end.tv_nsec = ns;
|
|
tm_end.tv_sec = 0;
|
|
tm_end = timespec64_add(tm_aux, tm_end);
|
|
|
|
/* do we have a full jiffie to wait? */
|
|
jiffies = NV_NSECS_TO_JIFFIES(ns);
|
|
|
|
if (jiffies)
|
|
{
|
|
//
|
|
// If we have at least one full jiffy to wait, give up
|
|
// up the CPU; since we may be rescheduled before
|
|
// the requested timeout has expired, loop until less
|
|
// than a jiffie of the desired delay remains.
|
|
//
|
|
set_current_state(TASK_INTERRUPTIBLE);
|
|
do
|
|
{
|
|
schedule_timeout(jiffies);
|
|
ktime_get_raw_ts64(&tm_aux);
|
|
if (nv_timer_less_than(&tm_aux, &tm_end))
|
|
{
|
|
tm_aux = timespec64_sub(tm_end, tm_aux);
|
|
ns = (NvU64) timespec64_to_ns(&tm_aux);
|
|
}
|
|
else
|
|
ns = 0;
|
|
} while ((jiffies = NV_NSECS_TO_JIFFIES(ns)) != 0);
|
|
}
|
|
|
|
if (ns > (NvU64) NSEC_PER_MSEC)
|
|
{
|
|
mdelay_safe_msec = ns / (NvU64) NSEC_PER_MSEC;
|
|
mdelay(mdelay_safe_msec);
|
|
ns %= (NvU64) NSEC_PER_MSEC;
|
|
}
|
|
if (ns)
|
|
{
|
|
ndelay(ns);
|
|
}
|
|
#ifdef NV_CHECK_DELAY_ACCURACY
|
|
ktime_get_raw_ts64(&tm_aux);
|
|
tm_aux = timespec64_sub(tm_aux, tm_start);
|
|
pr_info("NVRM: delay of %d msec results in actual delay of %lld.%09ld sec\n",
|
|
ms, tm_aux.tv_sec, tm_aux.tv_nsec);
|
|
#endif
|
|
return NV_OK;
|
|
}
|
|
|
|
#endif // __NV_TIME_H__
|