home *** CD-ROM | disk | FTP | other *** search
- #ifndef lint
- static char rcsid[] = "$Header: /usr/people/sam/tiff/libtiff/RCS/tif_machdep.c,v 1.2 92/03/06 11:52:42 sam Exp $";
- #endif
-
- /*
- * Copyright (c) 1992 Sam Leffler
- * Copyright (c) 1992 Silicon Graphics, Inc.
- *
- * Permission to use, copy, modify, distribute, and sell this software and
- * its documentation for any purpose is hereby granted without fee, provided
- * that (i) the above copyright notices and this permission notice appear in
- * all copies of the software and related documentation, and (ii) the names of
- * Sam Leffler and Silicon Graphics may not be used in any advertising or
- * publicity relating to the software without the specific, prior written
- * permission of Sam Leffler and Silicon Graphics.
- *
- * THE SOFTWARE IS PROVIDED "AS-IS" AND WITHOUT WARRANTY OF ANY KIND,
- * EXPRESS, IMPLIED OR OTHERWISE, INCLUDING WITHOUT LIMITATION, ANY
- * WARRANTY OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.
- *
- * IN NO EVENT SHALL SAM LEFFLER OR SILICON GRAPHICS BE LIABLE FOR
- * ANY SPECIAL, INCIDENTAL, INDIRECT OR CONSEQUENTIAL DAMAGES OF ANY KIND,
- * OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS,
- * WHETHER OR NOT ADVISED OF THE POSSIBILITY OF DAMAGE, AND ON ANY THEORY OF
- * LIABILITY, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE
- * OF THIS SOFTWARE.
- */
-
- /*
- * TIFF Library Machine Dependent Routines.
- */
- #include "tiffioP.h"
-
- #ifdef tahoe
- typedef struct ieeedouble {
- u_long sign : 1,
- exp : 11,
- mant : 20;
- u_long mant2;
- } ieeedouble;
- typedef struct ieeefloat {
- u_long sign : 1,
- exp : 8,
- mant : 23;
- } ieeefloat;
-
- typedef struct {
- u_long sign : 1,
- exp : 8,
- mant : 23;
- u_long mant2;
- } nativedouble;
- typedef struct {
- u_long sign : 1,
- exp : 8,
- mant : 23;
- } nativefloat;
- /*
- * Beware, over/under-flow in conversions will
- * result in garbage values -- handling it would
- * require a subroutine call or lots more code.
- */
- #define NATIVE2IEEEFLOAT(fp) { \
- if ((fp)->native.exp) \
- (fp)->ieee.exp = (fp)->native.exp - 129 + 127; /* alter bias */\
- }
- #define IEEEFLOAT2NATIVE(fp) { \
- if ((fp)->ieee.exp) \
- (fp)->native.exp = (fp)->ieee.exp - 127 + 129; /* alter bias */\
- }
- #define IEEEDOUBLE2NATIVE(dp) { \
- if ((dp)->native.exp = (dp)->ieee.exp) \
- (dp)->native.exp += -1023 + 129; \
- (dp)->native.mant = ((dp)->ieee.mant<<3)|((dp)->native.mant2>>29); \
- (dp)->native.mant2 <<= 3; \
- }
- #endif /* tahoe */
-
- #ifdef vax
- typedef struct ieeedouble {
- u_long mant : 20,
- exp : 11,
- sign : 1;
- u_long mant2;
- } ieeedouble;
- typedef struct ieeefloat {
- u_long mant : 23,
- exp : 8,
- sign : 1;
- } ieeefloat;
-
- typedef struct {
- u_long mant1 : 7,
- exp : 8,
- sign : 1,
- mant2 : 16;
- u_long mant3;
- } nativedouble;
- typedef struct {
- u_long mant1 : 7,
- exp : 8,
- sign : 1,
- mant2 : 16;
- } nativefloat;
- /*
- * Beware, these do not handle over/under-flow
- * during conversion from ieee to native format.
- */
- #define NATIVE2IEEEFLOAT(fp) { \
- float_t t; \
- if (t.ieee.exp = (fp)->native.exp) \
- t.ieee.exp += -129 + 127; \
- t.ieee.sign = (fp)->native.sign; \
- t.ieee.mant = ((fp)->native.mant1<<16)|(fp)->native.mant2; \
- *(fp) = t; \
- }
- #define IEEEFLOAT2NATIVE(fp) { \
- float_t t; int v = (fp)->ieee.exp; \
- if (v) v += -127 + 129; /* alter bias of exponent */\
- t.native.exp = v; /* implicit truncation of exponent */\
- t.native.sign = (fp)->ieee.sign; \
- v = (fp)->ieee.mant; \
- t.native.mant1 = v >> 16; \
- t.native.mant2 = v;\
- *(fp) = t; \
- }
- #define IEEEDOUBLE2NATIVE(dp) { \
- double_t t; int v = (dp)->ieee.exp; \
- if (v) v += -1023 + 129; /* if can alter bias of exponent */\
- t.native.exp = v; /* implicit truncation of exponent */\
- v = (dp)->ieee.mant; \
- t.native.sign = (dp)->ieee.sign; \
- t.native.mant1 = v >> 16; \
- t.native.mant2 = v;\
- t.native.mant3 = (dp)->mant2; \
- *(dp) = t; \
- }
- #endif /* vax */
-
- #if !HAVE_IEEEFP
- #if !defined(IEEEFLOAT2NATIVE) || !defined(NATIVE2IEEEFLOAT)
- "Help, you've configured the library to not have IEEE floating point,\
- but not defined how to convert between IEEE and native formats!"
- #endif
-
- /*
- * These unions are used during floating point
- * conversions. The above macros define the
- * conversion operations.
- */
- typedef union {
- ieeedouble ieee;
- nativedouble native;
- char b[8];
- double d;
- } double_t;
-
- typedef union {
- ieeefloat ieee;
- nativefloat native;
- char b[4];
- float f;
- } float_t;
-
- TIFFCvtIEEEFloatToNative(tif, n, f)
- TIFF *tif;
- u_int n;
- float *f;
- {
- float_t *fp = (float_t *)f;
-
- while (n-- > 0) {
- IEEEFLOAT2NATIVE(fp);
- fp++;
- }
- }
-
- TIFFCvtNativeToIEEEFloat(tif, n, f)
- TIFF *tif;
- u_int n;
- float *f;
- {
- float_t *fp = (float_t *)f;
-
- while (n-- > 0) {
- NATIVE2IEEEFLOAT(fp);
- fp++;
- }
- }
- #endif
-