413 lines
12 KiB
C
413 lines
12 KiB
C
|
/*
|
||
|
* ntp_types.h - defines how int32 and u_int32 are treated.
|
||
|
* For 64 bit systems like the DEC Alpha, they have to be defined
|
||
|
* as int and u_int.
|
||
|
* For 32 bit systems, define them as long and u_long
|
||
|
*/
|
||
|
#define SIZEOF_INT 4
|
||
|
|
||
|
/*
|
||
|
* Set up for prototyping
|
||
|
*/
|
||
|
#ifndef P
|
||
|
#if defined(__STDC__) || defined(HAVE_PROTOTYPES)
|
||
|
#define P(x) x
|
||
|
#else /* not __STDC__ and not HAVE_PROTOTYPES */
|
||
|
#define P(x) ()
|
||
|
#endif /* not __STDC__ and HAVE_PROTOTYPES */
|
||
|
#endif /* P */
|
||
|
|
||
|
/*
|
||
|
* VMS DECC (v4.1), {u_char,u_short,u_long} are only in SOCKET.H,
|
||
|
* and u_int isn't defined anywhere
|
||
|
*/
|
||
|
#if defined(VMS)
|
||
|
#include <socket.h>
|
||
|
typedef unsigned int u_int;
|
||
|
/*
|
||
|
* Note: VMS DECC has long == int (even on __alpha),
|
||
|
* so the distinction below doesn't matter
|
||
|
*/
|
||
|
#endif /* VMS */
|
||
|
|
||
|
#if (SIZEOF_INT == 4)
|
||
|
# ifndef int32
|
||
|
# define int32 int
|
||
|
# endif
|
||
|
# ifndef u_int32
|
||
|
# define u_int32 unsigned int
|
||
|
# endif
|
||
|
#else /* not sizeof(int) == 4 */
|
||
|
# if (SIZEOF_LONG == 4)
|
||
|
# else /* not sizeof(long) == 4 */
|
||
|
# ifndef int32
|
||
|
# define int32 long
|
||
|
# endif
|
||
|
# ifndef u_int32
|
||
|
# define u_int32 unsigned long
|
||
|
# endif
|
||
|
# endif /* not sizeof(long) == 4 */
|
||
|
# include "Bletch: what's 32 bits on this machine?"
|
||
|
#endif /* not sizeof(int) == 4 */
|
||
|
|
||
|
typedef unsigned short associd_t; /* association ID */
|
||
|
typedef u_int32 keyid_t; /* cryptographic key ID */
|
||
|
typedef u_int32 tstamp_t; /* NTP seconds timestamp */
|
||
|
|
||
|
/*
|
||
|
* NTP uses two fixed point formats. The first (l_fp) is the "long"
|
||
|
* format and is 64 bits long with the decimal between bits 31 and 32.
|
||
|
* This is used for time stamps in the NTP packet header (in network
|
||
|
* byte order) and for internal computations of offsets (in local host
|
||
|
* byte order). We use the same structure for both signed and unsigned
|
||
|
* values, which is a big hack but saves rewriting all the operators
|
||
|
* twice. Just to confuse this, we also sometimes just carry the
|
||
|
* fractional part in calculations, in both signed and unsigned forms.
|
||
|
* Anyway, an l_fp looks like:
|
||
|
*
|
||
|
* 0 1 2 3
|
||
|
* 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
|
||
|
* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|
||
|
* | Integral Part |
|
||
|
* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|
||
|
* | Fractional Part |
|
||
|
* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|
||
|
*
|
||
|
*/
|
||
|
typedef struct {
|
||
|
union {
|
||
|
u_int32 Xl_ui;
|
||
|
int32 Xl_i;
|
||
|
} Ul_i;
|
||
|
union {
|
||
|
u_int32 Xl_uf;
|
||
|
int32 Xl_f;
|
||
|
} Ul_f;
|
||
|
} l_fp;
|
||
|
|
||
|
#define l_ui Ul_i.Xl_ui /* unsigned integral part */
|
||
|
#define l_i Ul_i.Xl_i /* signed integral part */
|
||
|
#define l_uf Ul_f.Xl_uf /* unsigned fractional part */
|
||
|
#define l_f Ul_f.Xl_f /* signed fractional part */
|
||
|
|
||
|
/*
|
||
|
* Fractional precision (of an l_fp) is actually the number of
|
||
|
* bits in a long.
|
||
|
*/
|
||
|
#define FRACTION_PREC (32)
|
||
|
|
||
|
|
||
|
/*
|
||
|
* The second fixed point format is 32 bits, with the decimal between
|
||
|
* bits 15 and 16. There is a signed version (s_fp) and an unsigned
|
||
|
* version (u_fp). This is used to represent synchronizing distance
|
||
|
* and synchronizing dispersion in the NTP packet header (again, in
|
||
|
* network byte order) and internally to hold both distance and
|
||
|
* dispersion values (in local byte order). In network byte order
|
||
|
* it looks like:
|
||
|
*
|
||
|
* 0 1 2 3
|
||
|
* 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
|
||
|
* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|
||
|
* | Integer Part | Fraction Part |
|
||
|
* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|
||
|
*
|
||
|
*/
|
||
|
typedef int32 s_fp;
|
||
|
typedef u_int32 u_fp;
|
||
|
|
||
|
/*
|
||
|
* A unit second in fp format. Actually 2**(half_the_bits_in_a_long)
|
||
|
*/
|
||
|
#define FP_SECOND (0x10000)
|
||
|
|
||
|
/*
|
||
|
* Byte order conversions
|
||
|
*/
|
||
|
#define HTONS_FP(x) (htonl(x))
|
||
|
#define HTONL_FP(h, n) do { (n)->l_ui = htonl((h)->l_ui); \
|
||
|
(n)->l_uf = htonl((h)->l_uf); } while (0)
|
||
|
#define NTOHS_FP(x) (ntohl(x))
|
||
|
#define NTOHL_FP(n, h) do { (h)->l_ui = ntohl((n)->l_ui); \
|
||
|
(h)->l_uf = ntohl((n)->l_uf); } while (0)
|
||
|
#define NTOHL_MFP(ni, nf, hi, hf) \
|
||
|
do { (hi) = ntohl(ni); (hf) = ntohl(nf); } while (0)
|
||
|
#define HTONL_MFP(hi, hf, ni, nf) \
|
||
|
do { (ni) = ntohl(hi); (nf) = ntohl(hf); } while (0)
|
||
|
|
||
|
/* funny ones. Converts ts fractions to net order ts */
|
||
|
#define HTONL_UF(uf, nts) \
|
||
|
do { (nts)->l_ui = 0; (nts)->l_uf = htonl(uf); } while (0)
|
||
|
#define HTONL_F(f, nts) do { (nts)->l_uf = htonl(f); \
|
||
|
if ((f) & 0x80000000) \
|
||
|
(nts)->l_i = -1; \
|
||
|
else \
|
||
|
(nts)->l_i = 0; \
|
||
|
} while (0)
|
||
|
|
||
|
/*
|
||
|
* Conversions between the two fixed point types
|
||
|
*/
|
||
|
#define MFPTOFP(x_i, x_f) (((x_i) >= 0x00010000) ? 0x7fffffff : \
|
||
|
(((x_i) <= -0x00010000) ? 0x80000000 : \
|
||
|
(((x_i)<<16) | (((x_f)>>16)&0xffff))))
|
||
|
#define LFPTOFP(v) MFPTOFP((v)->l_i, (v)->l_f)
|
||
|
|
||
|
#define UFPTOLFP(x, v) ((v)->l_ui = (u_fp)(x)>>16, (v)->l_uf = (x)<<16)
|
||
|
#define FPTOLFP(x, v) (UFPTOLFP((x), (v)), (x) < 0 ? (v)->l_ui -= 0x10000 : 0)
|
||
|
|
||
|
#define MAXLFP(v) ((v)->l_ui = 0x7fffffff, (v)->l_uf = 0xffffffff)
|
||
|
#define MINLFP(v) ((v)->l_ui = 0x80000000, (v)->l_uf = 0)
|
||
|
|
||
|
/*
|
||
|
* Primitive operations on long fixed point values. If these are
|
||
|
* reminiscent of assembler op codes it's only because some may
|
||
|
* be replaced by inline assembler for particular machines someday.
|
||
|
* These are the (kind of inefficient) run-anywhere versions.
|
||
|
*/
|
||
|
#define M_NEG(v_i, v_f) /* v = -v */ \
|
||
|
do { \
|
||
|
if ((v_f) == 0) \
|
||
|
(v_i) = -((s_fp)(v_i)); \
|
||
|
else { \
|
||
|
(v_f) = -((s_fp)(v_f)); \
|
||
|
(v_i) = ~(v_i); \
|
||
|
} \
|
||
|
} while(0)
|
||
|
|
||
|
#define M_NEGM(r_i, r_f, a_i, a_f) /* r = -a */ \
|
||
|
do { \
|
||
|
if ((a_f) == 0) { \
|
||
|
(r_f) = 0; \
|
||
|
(r_i) = -(a_i); \
|
||
|
} else { \
|
||
|
(r_f) = -(a_f); \
|
||
|
(r_i) = ~(a_i); \
|
||
|
} \
|
||
|
} while(0)
|
||
|
|
||
|
#define M_ADD(r_i, r_f, a_i, a_f) /* r += a */ \
|
||
|
do { \
|
||
|
register u_int32 lo_tmp; \
|
||
|
register u_int32 hi_tmp; \
|
||
|
\
|
||
|
lo_tmp = ((r_f) & 0xffff) + ((a_f) & 0xffff); \
|
||
|
hi_tmp = (((r_f) >> 16) & 0xffff) + (((a_f) >> 16) & 0xffff); \
|
||
|
if (lo_tmp & 0x10000) \
|
||
|
hi_tmp++; \
|
||
|
(r_f) = ((hi_tmp & 0xffff) << 16) | (lo_tmp & 0xffff); \
|
||
|
\
|
||
|
(r_i) += (a_i); \
|
||
|
if (hi_tmp & 0x10000) \
|
||
|
(r_i)++; \
|
||
|
} while (0)
|
||
|
|
||
|
#define M_ADD3(r_ovr, r_i, r_f, a_ovr, a_i, a_f) /* r += a, three word */ \
|
||
|
do { \
|
||
|
register u_int32 lo_tmp; \
|
||
|
register u_int32 hi_tmp; \
|
||
|
\
|
||
|
lo_tmp = ((r_f) & 0xffff) + ((a_f) & 0xffff); \
|
||
|
hi_tmp = (((r_f) >> 16) & 0xffff) + (((a_f) >> 16) & 0xffff); \
|
||
|
if (lo_tmp & 0x10000) \
|
||
|
hi_tmp++; \
|
||
|
(r_f) = ((hi_tmp & 0xffff) << 16) | (lo_tmp & 0xffff); \
|
||
|
\
|
||
|
lo_tmp = ((r_i) & 0xffff) + ((a_i) & 0xffff); \
|
||
|
if (hi_tmp & 0x10000) \
|
||
|
lo_tmp++; \
|
||
|
hi_tmp = (((r_i) >> 16) & 0xffff) + (((a_i) >> 16) & 0xffff); \
|
||
|
if (lo_tmp & 0x10000) \
|
||
|
hi_tmp++; \
|
||
|
(r_i) = ((hi_tmp & 0xffff) << 16) | (lo_tmp & 0xffff); \
|
||
|
\
|
||
|
(r_ovr) += (a_ovr); \
|
||
|
if (hi_tmp & 0x10000) \
|
||
|
(r_ovr)++; \
|
||
|
} while (0)
|
||
|
|
||
|
#define M_SUB(r_i, r_f, a_i, a_f) /* r -= a */ \
|
||
|
do { \
|
||
|
register u_int32 lo_tmp; \
|
||
|
register u_int32 hi_tmp; \
|
||
|
\
|
||
|
if ((a_f) == 0) { \
|
||
|
(r_i) -= (a_i); \
|
||
|
} else { \
|
||
|
lo_tmp = ((r_f) & 0xffff) + ((-((s_fp)(a_f))) & 0xffff); \
|
||
|
hi_tmp = (((r_f) >> 16) & 0xffff) \
|
||
|
+ (((-((s_fp)(a_f))) >> 16) & 0xffff); \
|
||
|
if (lo_tmp & 0x10000) \
|
||
|
hi_tmp++; \
|
||
|
(r_f) = ((hi_tmp & 0xffff) << 16) | (lo_tmp & 0xffff); \
|
||
|
\
|
||
|
(r_i) += ~(a_i); \
|
||
|
if (hi_tmp & 0x10000) \
|
||
|
(r_i)++; \
|
||
|
} \
|
||
|
} while (0)
|
||
|
|
||
|
#define M_RSHIFTU(v_i, v_f) /* v >>= 1, v is unsigned */ \
|
||
|
do { \
|
||
|
(v_f) = (u_int32)(v_f) >> 1; \
|
||
|
if ((v_i) & 01) \
|
||
|
(v_f) |= 0x80000000; \
|
||
|
(v_i) = (u_int32)(v_i) >> 1; \
|
||
|
} while (0)
|
||
|
|
||
|
#define M_RSHIFT(v_i, v_f) /* v >>= 1, v is signed */ \
|
||
|
do { \
|
||
|
(v_f) = (u_int32)(v_f) >> 1; \
|
||
|
if ((v_i) & 01) \
|
||
|
(v_f) |= 0x80000000; \
|
||
|
if ((v_i) & 0x80000000) \
|
||
|
(v_i) = ((v_i) >> 1) | 0x80000000; \
|
||
|
else \
|
||
|
(v_i) = (v_i) >> 1; \
|
||
|
} while (0)
|
||
|
|
||
|
#define M_LSHIFT(v_i, v_f) /* v <<= 1 */ \
|
||
|
do { \
|
||
|
(v_i) <<= 1; \
|
||
|
if ((v_f) & 0x80000000) \
|
||
|
(v_i) |= 0x1; \
|
||
|
(v_f) <<= 1; \
|
||
|
} while (0)
|
||
|
|
||
|
#define M_LSHIFT3(v_ovr, v_i, v_f) /* v <<= 1, with overflow */ \
|
||
|
do { \
|
||
|
(v_ovr) <<= 1; \
|
||
|
if ((v_i) & 0x80000000) \
|
||
|
(v_ovr) |= 0x1; \
|
||
|
(v_i) <<= 1; \
|
||
|
if ((v_f) & 0x80000000) \
|
||
|
(v_i) |= 0x1; \
|
||
|
(v_f) <<= 1; \
|
||
|
} while (0)
|
||
|
|
||
|
#define M_ADDUF(r_i, r_f, uf) /* r += uf, uf is u_int32 fraction */ \
|
||
|
M_ADD((r_i), (r_f), 0, (uf)) /* let optimizer worry about it */
|
||
|
|
||
|
#define M_SUBUF(r_i, r_f, uf) /* r -= uf, uf is u_int32 fraction */ \
|
||
|
M_SUB((r_i), (r_f), 0, (uf)) /* let optimizer worry about it */
|
||
|
|
||
|
#define M_ADDF(r_i, r_f, f) /* r += f, f is a int32 fraction */ \
|
||
|
do { \
|
||
|
if ((f) > 0) \
|
||
|
M_ADD((r_i), (r_f), 0, (f)); \
|
||
|
else if ((f) < 0) \
|
||
|
M_ADD((r_i), (r_f), (-1), (f));\
|
||
|
} while(0)
|
||
|
|
||
|
#define M_ISNEG(v_i, v_f) /* v < 0 */ \
|
||
|
(((v_i) & 0x80000000) != 0)
|
||
|
|
||
|
#define M_ISHIS(a_i, a_f, b_i, b_f) /* a >= b unsigned */ \
|
||
|
(((u_int32)(a_i)) > ((u_int32)(b_i)) || \
|
||
|
((a_i) == (b_i) && ((u_int32)(a_f)) >= ((u_int32)(b_f))))
|
||
|
|
||
|
#define M_ISGEQ(a_i, a_f, b_i, b_f) /* a >= b signed */ \
|
||
|
(((int32)(a_i)) > ((int32)(b_i)) || \
|
||
|
((a_i) == (b_i) && ((u_int32)(a_f)) >= ((u_int32)(b_f))))
|
||
|
|
||
|
#define M_ISEQU(a_i, a_f, b_i, b_f) /* a == b unsigned */ \
|
||
|
((a_i) == (b_i) && (a_f) == (b_f))
|
||
|
|
||
|
/*
|
||
|
* Operations on the long fp format
|
||
|
*/
|
||
|
#define L_ADD(r, a) M_ADD((r)->l_ui, (r)->l_uf, (a)->l_ui, (a)->l_uf)
|
||
|
#define L_SUB(r, a) M_SUB((r)->l_ui, (r)->l_uf, (a)->l_ui, (a)->l_uf)
|
||
|
#define L_NEG(v) M_NEG((v)->l_ui, (v)->l_uf)
|
||
|
#define L_ADDUF(r, uf) M_ADDUF((r)->l_ui, (r)->l_uf, (uf))
|
||
|
#define L_SUBUF(r, uf) M_SUBUF((r)->l_ui, (r)->l_uf, (uf))
|
||
|
#define L_ADDF(r, f) M_ADDF((r)->l_ui, (r)->l_uf, (f))
|
||
|
#define L_RSHIFT(v) M_RSHIFT((v)->l_i, (v)->l_uf)
|
||
|
#define L_RSHIFTU(v) M_RSHIFT((v)->l_ui, (v)->l_uf)
|
||
|
#define L_LSHIFT(v) M_LSHIFT((v)->l_ui, (v)->l_uf)
|
||
|
#define L_CLR(v) ((v)->l_ui = (v)->l_uf = 0)
|
||
|
|
||
|
#define L_ISNEG(v) (((v)->l_ui & 0x80000000) != 0)
|
||
|
#define L_ISZERO(v) ((v)->l_ui == 0 && (v)->l_uf == 0)
|
||
|
#define L_ISHIS(a, b) ((a)->l_ui > (b)->l_ui || \
|
||
|
((a)->l_ui == (b)->l_ui && (a)->l_uf >= (b)->l_uf))
|
||
|
#define L_ISGEQ(a, b) ((a)->l_i > (b)->l_i || \
|
||
|
((a)->l_i == (b)->l_i && (a)->l_uf >= (b)->l_uf))
|
||
|
#define L_ISEQU(a, b) M_ISEQU((a)->l_ui, (a)->l_uf, (b)->l_ui, (b)->l_uf)
|
||
|
|
||
|
/*
|
||
|
* s_fp/double and u_fp/double conversions
|
||
|
*/
|
||
|
#define FRIC 65536. /* 2^16 as a double */
|
||
|
#define DTOFP(r) ((s_fp)((r) * FRIC))
|
||
|
#define DTOUFP(r) ((u_fp)((r) * FRIC))
|
||
|
#define FPTOD(r) ((double)(r) / FRIC)
|
||
|
|
||
|
/*
|
||
|
* l_fp/double conversions
|
||
|
*/
|
||
|
#define FRAC 4294967296. /* 2^32 as a double */
|
||
|
#define M_DTOLFP(d, r_i, r_uf) /* double to l_fp */ \
|
||
|
do { \
|
||
|
register double d_tmp; \
|
||
|
\
|
||
|
d_tmp = (d); \
|
||
|
if (d_tmp < 0) { \
|
||
|
d_tmp = -d_tmp; \
|
||
|
(r_i) = (int32)(d_tmp); \
|
||
|
(r_uf) = (u_int32)(((d_tmp) - (double)(r_i)) * FRAC); \
|
||
|
M_NEG((r_i), (r_uf)); \
|
||
|
} else { \
|
||
|
(r_i) = (int32)(d_tmp); \
|
||
|
(r_uf) = (u_int32)(((d_tmp) - (double)(r_i)) * FRAC); \
|
||
|
} \
|
||
|
} while (0)
|
||
|
#define M_LFPTOD(r_i, r_uf, d) /* l_fp to double */ \
|
||
|
do { \
|
||
|
register l_fp l_tmp; \
|
||
|
\
|
||
|
l_tmp.l_i = (r_i); \
|
||
|
l_tmp.l_f = (r_uf); \
|
||
|
if (l_tmp.l_i < 0) { \
|
||
|
M_NEG(l_tmp.l_i, l_tmp.l_uf); \
|
||
|
(d) = -((double)l_tmp.l_i + ((double)l_tmp.l_uf) / FRAC); \
|
||
|
} else { \
|
||
|
(d) = (double)l_tmp.l_i + ((double)l_tmp.l_uf) / FRAC; \
|
||
|
} \
|
||
|
} while (0)
|
||
|
#define DTOLFP(d, v) M_DTOLFP((d), (v)->l_ui, (v)->l_uf)
|
||
|
#define LFPTOD(v, d) M_LFPTOD((v)->l_ui, (v)->l_uf, (d))
|
||
|
|
||
|
/*
|
||
|
* Prototypes
|
||
|
*/
|
||
|
#if 0
|
||
|
extern char * dofptoa P((u_fp, int, short, int));
|
||
|
extern char * dolfptoa P((u_long, u_long, int, short, int));
|
||
|
#endif
|
||
|
|
||
|
extern int atolfp P((const char *, l_fp *));
|
||
|
extern int buftvtots P((const char *, l_fp *));
|
||
|
extern char * fptoa P((s_fp, short));
|
||
|
extern char * fptoms P((s_fp, short));
|
||
|
extern int hextolfp P((const char *, l_fp *));
|
||
|
extern void gpstolfp P((int, int, unsigned long, l_fp *));
|
||
|
extern int mstolfp P((const char *, l_fp *));
|
||
|
extern char * prettydate P((l_fp *));
|
||
|
extern char * gmprettydate P((l_fp *));
|
||
|
extern char * uglydate P((l_fp *));
|
||
|
extern void mfp_mul P((int32 *, u_int32 *, int32, u_int32, int32, u_int32));
|
||
|
|
||
|
extern void get_systime P((l_fp *));
|
||
|
extern int step_systime P((double));
|
||
|
extern int adj_systime P((double));
|
||
|
|
||
|
#define lfptoa(_fpv, _ndec) mfptoa((_fpv)->l_ui, (_fpv)->l_uf, (_ndec))
|
||
|
#define lfptoms(_fpv, _ndec) mfptoms((_fpv)->l_ui, (_fpv)->l_uf, (_ndec))
|
||
|
|
||
|
#define ufptoa(_fpv, _ndec) dofptoa((_fpv), 0, (_ndec), 0)
|
||
|
#define ufptoms(_fpv, _ndec) dofptoa((_fpv), 0, (_ndec), 1)
|
||
|
#define ulfptoa(_fpv, _ndec) dolfptoa((_fpv)->l_ui, (_fpv)->l_uf, 0, (_ndec), 0)
|
||
|
#define ulfptoms(_fpv, _ndec) dolfptoa((_fpv)->l_ui, (_fpv)->l_uf, 0, (_ndec), 1)
|
||
|
#define umfptoa(_fpi, _fpf, _ndec) dolfptoa((_fpi), (_fpf), 0, (_ndec), 0)
|