opnsense-src/sys/compat/linux/linux_vdso_gtod.inc
Dmitry Chagin 9931033bbf linux(4); Almost complete the vDSO.
The vDSO (virtual dynamic shared object) is a small shared library that the
kernel maps R/O into the address space of all Linux processes on image
activation. The vDSO is a fully formed ELF image, shared by all processes
with the same ABI, has no process private data.

The primary purpose of the vDSO:
- non-executable stack, signal trampolines not copied to the stack;
- signal trampolines unwind, mandatory for the NPTL;
- to avoid contex-switch overhead frequently used system calls can be
  implemented in the vDSO: for now gettimeofday, clock_gettime.

The first two have been implemented, so add the implementation of system
calls.

System calls implemenation based on a native timekeeping code with some
limitations:
- ifunc can't be used, as vDSO r/o mapped to the process VA and rtld
  can't relocate symbols;
- reading HPET memory is not implemented for now (TODO).

In case on any error vDSO system calls fallback to the kernel system
calls. For unimplemented vDSO system calls added prototypes which call
corresponding kernel system call.

Tested by:		trasz (arm64)
Differential revision:  https://reviews.freebsd.org/D30900
MFC after:              2 weeks
2021-07-20 10:01:18 +03:00

337 lines
7.9 KiB
C++

/*-
* SPDX-License-Identifier: BSD-2-Clause-FreeBSD
*
* Copyright (c) 2012 Konstantin Belousov <kib@FreeBSD.org>
* Copyright (c) 2021 Dmitry Chagin <dchagin@FreeBSD.org>
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*/
static int
__vdso_native_to_linux_timespec(struct l_timespec *lts,
struct timespec *nts)
{
#ifdef COMPAT_LINUX32
if (nts->tv_sec > INT_MAX || nts->tv_sec < INT_MIN)
return (LINUX_EOVERFLOW);
#endif
lts->tv_sec = nts->tv_sec;
lts->tv_nsec = nts->tv_nsec;
return (0);
}
static int
__vdso_native_to_linux_timeval(l_timeval *ltv,
struct timeval *ntv)
{
#ifdef COMPAT_LINUX32
if (ntv->tv_sec > INT_MAX || ntv->tv_sec < INT_MIN)
return (LINUX_EOVERFLOW);
#endif
ltv->tv_sec = ntv->tv_sec;
ltv->tv_usec = ntv->tv_usec;
return (0);
}
#if defined(__i386__) || (defined(__amd64__) && defined(COMPAT_LINUX32))
static int
__vdso_native_to_linux_timespec64(struct l_timespec64 *lts,
struct timespec *nts)
{
lts->tv_sec = nts->tv_sec;
lts->tv_nsec = nts->tv_nsec;
return (0);
}
#endif
static int
__vdso_linux_to_native_clockid(clockid_t *n, clockid_t l)
{
switch (l) {
case LINUX_CLOCK_REALTIME:
*n = CLOCK_REALTIME;
break;
case LINUX_CLOCK_MONOTONIC:
*n = CLOCK_MONOTONIC;
break;
case LINUX_CLOCK_REALTIME_COARSE:
*n = CLOCK_REALTIME_FAST;
break;
case LINUX_CLOCK_MONOTONIC_COARSE:
case LINUX_CLOCK_MONOTONIC_RAW:
*n = CLOCK_MONOTONIC_FAST;
break;
case LINUX_CLOCK_BOOTTIME:
*n = CLOCK_UPTIME;
break;
default:
return (LINUX_EINVAL);
}
return (0);
}
/*
* The code below adapted from
* lib/libc/sys/__vdso_gettimeofday.c
*/
static inline void
__vdso_gettimekeep(struct vdso_timekeep **tk)
{
*tk = (struct vdso_timekeep *)kern_timekeep_base;
}
static int
tc_delta(const struct vdso_timehands *th, u_int *delta)
{
int error;
u_int tc;
error = __vdso_gettc(th, &tc);
if (error == 0)
*delta = (tc - th->th_offset_count) & th->th_counter_mask;
return (error);
}
/*
* Calculate the absolute or boot-relative time from the
* machine-specific fast timecounter and the published timehands
* structure read from the shared page.
*
* The lockless reading scheme is similar to the one used to read the
* in-kernel timehands, see sys/kern/kern_tc.c:binuptime(). This code
* is based on the kernel implementation.
*/
static int
freebsd_binuptime(struct bintime *bt, struct vdso_timekeep *tk, bool abs)
{
struct vdso_timehands *th;
uint32_t curr, gen;
uint64_t scale, x;
u_int delta, scale_bits;
int error;
do {
if (!tk->tk_enabled)
return (ENOSYS);
curr = atomic_load_acq_32(&tk->tk_current);
th = &tk->tk_th[curr];
gen = atomic_load_acq_32(&th->th_gen);
*bt = th->th_offset;
error = tc_delta(th, &delta);
if (error == EAGAIN)
continue;
if (error != 0)
return (error);
scale = th->th_scale;
#ifdef _LP64
scale_bits = ffsl(scale);
#else
scale_bits = ffsll(scale);
#endif
if (__predict_false(scale_bits + fls(delta) > 63)) {
x = (scale >> 32) * delta;
scale &= 0xffffffff;
bt->sec += x >> 32;
bintime_addx(bt, x << 32);
}
bintime_addx(bt, scale * delta);
if (abs)
bintime_add(bt, &th->th_boottime);
/*
* Ensure that the load of th_offset is completed
* before the load of th_gen.
*/
atomic_thread_fence_acq();
} while (curr != tk->tk_current || gen == 0 || gen != th->th_gen);
return (0);
}
static int
freebsd_getnanouptime(struct bintime *bt, struct vdso_timekeep *tk)
{
struct vdso_timehands *th;
uint32_t curr, gen;
do {
if (!tk->tk_enabled)
return (ENOSYS);
curr = atomic_load_acq_32(&tk->tk_current);
th = &tk->tk_th[curr];
gen = atomic_load_acq_32(&th->th_gen);
*bt = th->th_offset;
/*
* Ensure that the load of th_offset is completed
* before the load of th_gen.
*/
atomic_thread_fence_acq();
} while (curr != tk->tk_current || gen == 0 || gen != th->th_gen);
return (0);
}
static int
freebsd_gettimeofday(struct timeval *tv, struct timezone *tz)
{
struct vdso_timekeep *tk;
struct bintime bt;
int error;
if (tz != NULL)
return (ENOSYS);
__vdso_gettimekeep(&tk);
if (tk == NULL)
return (ENOSYS);
if (tk->tk_ver != VDSO_TK_VER_CURR)
return (ENOSYS);
error = freebsd_binuptime(&bt, tk, true);
if (error == 0)
bintime2timeval(&bt, tv);
return (error);
}
static int
freebsd_clock_gettime(clockid_t clock_id, struct timespec *ts)
{
struct vdso_timekeep *tk;
struct bintime bt;
int error;
__vdso_gettimekeep(&tk);
if (tk == NULL)
return (ENOSYS);
if (tk->tk_ver != VDSO_TK_VER_CURR)
return (ENOSYS);
switch (clock_id) {
case CLOCK_REALTIME:
case CLOCK_REALTIME_PRECISE:
case CLOCK_REALTIME_FAST:
error = freebsd_binuptime(&bt, tk, true);
break;
case CLOCK_MONOTONIC:
case CLOCK_MONOTONIC_PRECISE:
case CLOCK_UPTIME:
case CLOCK_UPTIME_PRECISE:
error = freebsd_binuptime(&bt, tk, false);
break;
case CLOCK_MONOTONIC_FAST:
case CLOCK_UPTIME_FAST:
error = freebsd_getnanouptime(&bt, tk);
break;
default:
error = ENOSYS;
break;
}
if (error == 0)
bintime2timespec(&bt, ts);
return (error);
}
/*
* Linux vDSO interfaces
*
*/
int
__vdso_clock_gettime(clockid_t clock_id, struct l_timespec *lts)
{
struct timespec ts;
clockid_t which;
int error;
error = __vdso_linux_to_native_clockid(&which, clock_id);
if (error != 0)
return (__vdso_clock_gettime_fallback(clock_id, lts));
error = freebsd_clock_gettime(which, &ts);
if (error == 0)
return (-__vdso_native_to_linux_timespec(lts, &ts));
else
return (__vdso_clock_gettime_fallback(clock_id, lts));
}
int
__vdso_gettimeofday(l_timeval *ltv, struct timezone *tz)
{
struct timeval tv;
int error;
error = freebsd_gettimeofday(&tv, tz);
if (error != 0)
return (__vdso_gettimeofday_fallback(ltv, tz));
return (-__vdso_native_to_linux_timeval(ltv, &tv));
}
int
__vdso_clock_getres(clockid_t clock_id, struct l_timespec *lts)
{
return (__vdso_clock_getres_fallback(clock_id, lts));
}
#if defined(__i386__) || defined(COMPAT_LINUX32)
int
__vdso_clock_gettime64(clockid_t clock_id, struct l_timespec64 *lts)
{
struct timespec ts;
clockid_t which;
int error;
error = __vdso_linux_to_native_clockid(&which, clock_id);
if (error != 0)
return (__vdso_clock_gettime64_fallback(clock_id, lts));
error = freebsd_clock_gettime(which, &ts);
if (error == 0)
return(-__vdso_native_to_linux_timespec64(lts, &ts));
else
return(__vdso_clock_gettime64_fallback(clock_id, lts));
}
int clock_gettime64(clockid_t clock_id, struct l_timespec64 *lts)
__attribute__((weak, alias("__vdso_clock_gettime64")));
#endif
#if defined(__amd64__) && !defined(COMPAT_LINUX32)
int
__vdso_getcpu(uint32_t *cpu, uint32_t *node, void *cache)
{
return (__vdso_getcpu_fallback(cpu, node, cache));
}
#endif
#if defined(__i386__) || defined(__amd64__)
int
__vdso_time(long *tm)
{
return (__vdso_time_fallback(tm));
}
#endif