1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
|
// SPDX-License-Identifier: 0BSD
///////////////////////////////////////////////////////////////////////////////
//
/// \file mytime.c
/// \brief Time handling functions
//
// Author: Lasse Collin
//
///////////////////////////////////////////////////////////////////////////////
#include "private.h"
#if defined(MYTHREAD_VISTA) || defined(_MSC_VER)
// Nothing
#elif defined(HAVE_CLOCK_GETTIME) \
&& (!defined(__MINGW32__) || defined(MYTHREAD_POSIX))
# include <time.h>
#else
# include <sys/time.h>
#endif
uint64_t opt_flush_timeout = 0;
// start_time holds the time when the (de)compression was started.
// It's from mytime_now() and thus only useful for calculating relative
// time differences (elapsed time). start_time is initialized by calling
// mytime_set_start_time() and modified by mytime_sigtstp_handler().
//
// When mytime_sigtstp_handler() is used, start_time is made volatile.
// I'm not sure if that is really required since access to it is guarded
// by signals_block()/signals_unblock() since accessing an uint64_t isn't
// atomic on all systems. But since the variable isn't accessed very
// frequently making it volatile doesn't hurt.
#ifdef USE_SIGTSTP_HANDLER
static volatile uint64_t start_time;
#else
static uint64_t start_time;
#endif
static uint64_t next_flush;
/// \brief Get the current time as milliseconds
///
/// It's relative to some point but not necessarily to the UNIX Epoch.
static uint64_t
mytime_now(void)
{
#if defined(MYTHREAD_VISTA) || defined(_MSC_VER)
// Since there is no SIGALRM on Windows, this function gets
// called frequently when the progress indicator is in use.
// Progress indicator doesn't need high-resolution time.
// GetTickCount64() has very low overhead but needs at least WinVista.
//
// MinGW-w64 provides the POSIX functions clock_gettime() and
// gettimeofday() in a manner that allow xz to run on older
// than WinVista. If the threading method needs WinVista anyway,
// there's no reason to avoid a WinVista API here either.
return GetTickCount64();
#elif defined(HAVE_CLOCK_GETTIME) \
&& (!defined(__MINGW32__) || defined(MYTHREAD_POSIX))
// MinGW-w64: clock_gettime() is defined in winpthreads but we need
// nothing else from winpthreads (unless, for some odd reason, POSIX
// threading has been selected). By avoiding clock_gettime(), we
// avoid the dependency on libwinpthread-1.dll or the need to link
// against the static version. The downside is that the fallback
// method, gettimeofday(), doesn't provide monotonic time.
struct timespec tv;
# ifdef HAVE_CLOCK_MONOTONIC
// If CLOCK_MONOTONIC was available at compile time but for some
// reason isn't at runtime, fallback to CLOCK_REALTIME which
// according to POSIX is mandatory for all implementations.
static clockid_t clk_id = CLOCK_MONOTONIC;
while (clock_gettime(clk_id, &tv))
clk_id = CLOCK_REALTIME;
# else
clock_gettime(CLOCK_REALTIME, &tv);
# endif
return (uint64_t)tv.tv_sec * 1000 + (uint64_t)(tv.tv_nsec / 1000000);
#else
struct timeval tv;
gettimeofday(&tv, NULL);
return (uint64_t)tv.tv_sec * 1000 + (uint64_t)(tv.tv_usec / 1000);
#endif
}
#ifdef USE_SIGTSTP_HANDLER
extern void
mytime_sigtstp_handler(int sig lzma_attribute((__unused__)))
{
// Measure how long the process stays in the stopped state and add
// that amount to start_time. This way the the progress indicator
// won't count the stopped time as elapsed time and the estimated
// remaining time won't be confused by the time spent in the
// stopped state.
//
// FIXME? Is raising SIGSTOP the correct thing to do? POSIX.1-2017
// says that orphan processes shouldn't stop on SIGTSTP. So perhaps
// the most correct thing to do could be to revert to the default
// handler for SIGTSTP, unblock SIGTSTP, and then raise(SIGTSTP).
// It's quite a bit more complicated than just raising SIGSTOP though.
//
// The difference between raising SIGTSTP vs. SIGSTOP can be seen on
// the shell command line too by running "echo $?" after stopping
// a process but perhaps that doesn't matter.
const uint64_t t = mytime_now();
raise(SIGSTOP);
start_time += mytime_now() - t;
return;
}
#endif
extern void
mytime_set_start_time(void)
{
#ifdef USE_SIGTSTP_HANDLER
// Block the signals when accessing start_time so that we cannot
// end up with a garbage value. start_time is volatile but access
// to it isn't atomic at least on 32-bit systems.
signals_block();
#endif
start_time = mytime_now();
#ifdef USE_SIGTSTP_HANDLER
signals_unblock();
#endif
return;
}
extern uint64_t
mytime_get_elapsed(void)
{
#ifdef USE_SIGTSTP_HANDLER
signals_block();
#endif
const uint64_t t = mytime_now() - start_time;
#ifdef USE_SIGTSTP_HANDLER
signals_unblock();
#endif
return t;
}
extern void
mytime_set_flush_time(void)
{
next_flush = mytime_now() + opt_flush_timeout;
return;
}
extern int
mytime_get_flush_timeout(void)
{
if (opt_flush_timeout == 0 || opt_mode != MODE_COMPRESS)
return -1;
const uint64_t now = mytime_now();
if (now >= next_flush)
return 0;
const uint64_t remaining = next_flush - now;
return remaining > INT_MAX ? INT_MAX : (int)remaining;
}
|