blob: ebbda28698729539872ad79f62d61cd1a9d67bc0 (
plain)
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
|
/*
* This file and its contents are supplied under the terms of the
* Common Development and Distribution License ("CDDL"), version 1.0.
* You may only use this file in accordance with the terms of version
* 1.0 of the CDDL.
*
* A full copy of the text of the CDDL should have accompanied this
* source. A copy of the CDDL is also available via the Internet at
* http://www.illumos.org/license/CDDL.
*/
/*
* Copyright 2018 Joyent, Inc.
*/
#ifndef _LX_USERHZ_H
#define _LX_USERHZ_H
#ifdef __cplusplus
extern "C" {
#endif
/*
* Within the kernel, Linux implements an internal hz that they refer to as a
* "jiffy". Linux can be built with different hz, but on modern kernels
* it is frequently 250. However, Linux has a separate concept for the hz
* that is visible outside the kernel. This is called "USER_HZ" and is the
* value returned by 'sysconf(_SC_CLK_TCK)'. This is almost universally set to
* 100hz. Some (lazy) applications just hardcode 100hz instead of checking.
* To accommodate these broken applications, we always work with a USER_HZ of
* 100 and scale accordingly. See the Linux time(7) man page for a more
* detailed discussion of their behavior. See the comment in our
* uts/common/conf/param.c for a discussion of valid native hz values.
*
* There are a few interfaces which expose a clock_t to user-land and which
* need to be considered for USER_HZ adjustment.
* 1) The times(2) syscall. This is handled correctly.
* 2) The waitid(2) syscall passes a siginfo_t which contains si_stime and
* si_utime. Testing waitid(2) on various Linux distributions shows that the
* these fields are garbage. This aligns with the Linux waitid(2) man page,
* which describes the subset of the siginfo_t structure that is populated.
* Neither si_stime or si_utime are listed.
* 3) A sigaction(2) handler can pass a siginfo_t. This is only documented to
* occur when the sa_flags is SA_SIGINFO. The si_stime and si_utime are
* documented to only be populated when the signal is SIGCHLD. However,
* testing on Linux seems to show that these fields are not consistent
* with the corresponding times(2) data for the process, even for the
* SIGCHLD sigaction handler case.
* 4) Some fields in /proc/stat and /proc/pid/stat. See the Linux proc man
* page for references to sysconf(_SC_CLK_TCK).
*
* Although the siginfo_t si_stime and si_utime data for cases #2 and #3 is not
* consistent on Linux, we populate these fields correctly to be on the safe
* side.
*/
extern uint_t lx_hz_scale;
#define LX_USERHZ 100
#define HZ_TO_LX_USERHZ(x) ((x) / lx_hz_scale)
#ifdef __cplusplus
}
#endif
#endif /* _LX_USERHZ_H */
|
