The Zigbee ZCL Epoch is *after* the UN*X epoch, which means, given that all the epoch delta #defines in wsutil/epochs.h are unsigned, that the delta needs to be *added* to a time in seconds since the Zigbee ZCL Epoch to give a time in seconds since the UN*X epoch. This is the opposite operation for all the other epoch deltas in epochs.h, which refer to epoch absolute times *before* the UN*X epoch. Note the correct operation is done in the various Zigbee dissectors already; the problem is only in epan/proto.c Verified with the sample capture in #14138 Fixup 00e069e6cd9d72e79f207bbb5ddc6901c6e14420
83 lines
2.6 KiB
C
83 lines
2.6 KiB
C
/** @file
|
|
*
|
|
* Definitions of epoch values for various absolute time types.
|
|
*
|
|
* Wireshark - Network traffic analyzer
|
|
* By Gerald Combs <gerald@wireshark.org>
|
|
* Copyright 2006 Gerald Combs
|
|
*
|
|
* SPDX-License-Identifier: GPL-2.0-or-later
|
|
*/
|
|
|
|
#ifndef __EPOCHS_H__
|
|
#define __EPOCHS_H__
|
|
|
|
#include <glib.h>
|
|
|
|
/*
|
|
* Deltas between the epochs for various non-UN*X time stamp formats and
|
|
* the January 1, 1970, 00:00:00 (proleptic?) UTC epoch for the UN*X time
|
|
* stamp format.
|
|
*
|
|
* XXX - All of these are unsigned, but note that most of these epochs are
|
|
* before the UN*X epoch but one is afterwards, which changes whether the
|
|
* delta should be subtracted or added to a time in the other epoch to
|
|
* produce a time_t in the UN*X epoch. Perhaps they should be signed,
|
|
* or there should macros that do the arithmetic and return a time_t?
|
|
*/
|
|
|
|
/*
|
|
* 1900-01-01 00:00:00 (proleptic?) UTC.
|
|
* Used by a number of time formats.
|
|
*/
|
|
#define EPOCH_DELTA_1900_01_01_00_00_00_UTC 2208988800U
|
|
|
|
/*
|
|
* 1904-01-01 00:00:00 (proleptic?) UTC.
|
|
* Used in the classic Mac OS, and by formats, such as MPEG-4 Part 14 (MP4),
|
|
* which is based on Apple's QuickTime format.
|
|
*/
|
|
#define EPOCH_DELTA_1904_01_01_00_00_00_UTC 2082844800U
|
|
|
|
/*
|
|
* 1601-01-01 (proleptic Gregorian) 00:00:00 (proleptic?) UTC.
|
|
* The Windows NT epoch, used in a number of places, as it is
|
|
* the start of a 400 year Gregorian cycle.
|
|
*
|
|
* This is
|
|
*
|
|
* 369*365.25*24*60*60-(3*24*60*60+6*60*60)
|
|
*
|
|
* or equivalently,
|
|
*
|
|
* (89*4*365.25+(3*4+1)*365)*24*60*60
|
|
*
|
|
* 1970-1601 is 369; 365.25 is the average length of a year in days,
|
|
* including leap years.
|
|
*
|
|
* 369 = 4*92 + 1, so there are 92 groups of 4 consecutive years plus
|
|
* one leftover year, 1969, with 365 days.
|
|
*
|
|
* All but three of the groups of 4 consecutive years average 365.25 days
|
|
* per year, as they have one leap year in the group. However, 1700, 1800,
|
|
* and 1900 were not leap years, as, while they're all evenly divisible by 4,
|
|
* they're also evenly divisible by 100, but not evenly divisible by 400.
|
|
*
|
|
* So we have 89 groups of 4 consecutive years that average 365.25
|
|
* days per year, 3 groups of 4 consecutive years that average 365 days
|
|
* (as they lack a leap year), and one leftover year, 1969, that is
|
|
* 365 days long.
|
|
*/
|
|
#define EPOCH_DELTA_1601_01_01_00_00_00_UTC UINT64_C(11644473600)
|
|
|
|
/*
|
|
* 2000-01-01 00:00:00 UTC.
|
|
* Used by the Zigbee Zigbee Cluster Library protocol.
|
|
*
|
|
* This is *after* the UN*X epoch, so to convert from a time in seconds from
|
|
* this epoch this delta needs to be *added*, unlike the other values.
|
|
*/
|
|
#define EPOCH_DELTA_2000_01_01_00_00_00_UTC UINT64_C(((3*365 + 366)*7 + 2*365)*24*3600)
|
|
|
|
#endif /* __EPOCHS_H__ */
|