source: clfs-embedded/patches/gmp-4.2.4-branch_update-1.patch@ 4a4fe83

Last change on this file since 4a4fe83 was 06beb7a, checked in by Jim Gifford <clfs@…>, 16 years ago

Added: Patches
  • Property mode set to 100644
File size: 10.8 KB
RevLine 
[06beb7a]1Submitted By: Jim Gifford (jim at cross-lfs dot org)
2Date: 2009-01-03
3Initial Package Version: 4.2.4
4Origin: GMP Website
5Upstream Status: Fixed
6Description: See http://gmplib.org Website Under Curent Status
7
8diff -Naur gmp-4.2.4.orig/doc/gmp.texi gmp-4.2.4/doc/gmp.texi
9--- gmp-4.2.4.orig/doc/gmp.texi 2008-09-18 11:36:14.000000000 -0400
10+++ gmp-4.2.4/doc/gmp.texi 2009-01-03 13:48:39.498471376 -0500
11@@ -4849,9 +4849,12 @@
12 equal, zero otherwise. I.e., test if @var{op1} and @var{op2} are approximately
13 equal.
14
15-Caution: Currently only whole limbs are compared, and only in an exact
16-fashion. In the future values like 1000 and 0111 may be considered the same
17-to 3 bits (on the basis that their difference is that small).
18+Caution 1: All version of GMP up to version 4.2.4 compared just whole limbs,
19+meaning sometimes more than @var{op3} bits, sometimes fewer.
20+
21+Caution 2: This function will consider XXX11...111 and XX100...000 different,
22+even if ... is replaced by a semi-infinite number of bits. Such numbers are
23+really just one ulp off, and should be considered equal.
24 @end deftypefun
25
26 @deftypefun void mpf_reldiff (mpf_t @var{rop}, mpf_t @var{op1}, mpf_t @var{op2})
27diff -Naur gmp-4.2.4.orig/mpf/eq.c gmp-4.2.4/mpf/eq.c
28--- gmp-4.2.4.orig/mpf/eq.c 2007-08-30 14:31:40.000000000 -0400
29+++ gmp-4.2.4/mpf/eq.c 2009-01-03 13:48:39.498471376 -0500
30@@ -1,6 +1,6 @@
31 /* mpf_eq -- Compare two floats up to a specified bit #.
32
33-Copyright 1993, 1995, 1996, 2001, 2002 Free Software Foundation, Inc.
34+Copyright 1993, 1995, 1996, 2001, 2002, 2008 Free Software Foundation, Inc.
35
36 This file is part of the GNU MP Library.
37
38@@ -19,6 +19,7 @@
39
40 #include "gmp.h"
41 #include "gmp-impl.h"
42+#include "longlong.h"
43
44 int
45 mpf_eq (mpf_srcptr u, mpf_srcptr v, unsigned long int n_bits)
46@@ -26,6 +27,8 @@
47 mp_srcptr up, vp;
48 mp_size_t usize, vsize, size, i;
49 mp_exp_t uexp, vexp;
50+ mp_limb_t diff;
51+ int cnt;
52
53 uexp = u->_mp_exp;
54 vexp = v->_mp_exp;
55@@ -53,10 +56,8 @@
56 /* U and V have the same sign and are both non-zero. */
57
58 /* 2. Are the exponents different? */
59- if (uexp > vexp)
60- return 0; /* ??? handle (uexp = vexp + 1) */
61- if (vexp > uexp)
62- return 0; /* ??? handle (vexp = uexp + 1) */
63+ if (uexp != vexp)
64+ return 0;
65
66 usize = ABS (usize);
67 vsize = ABS (vsize);
68@@ -93,17 +94,26 @@
69 size = usize;
70 }
71
72- if (size > (n_bits + GMP_NUMB_BITS - 1) / GMP_NUMB_BITS)
73- size = (n_bits + GMP_NUMB_BITS - 1) / GMP_NUMB_BITS;
74+ up += usize; /* point just above most significant limb */
75+ vp += vsize; /* point just above most significant limb */
76
77- up += usize - size;
78- vp += vsize - size;
79+ count_leading_zeros (cnt, up[-1]);
80+ if ((vp[-1] >> (GMP_LIMB_BITS - 1 - cnt)) != 1)
81+ return 0; /* msb positions different */
82
83- for (i = size - 1; i >= 0; i--)
84+ n_bits += cnt - GMP_NAIL_BITS;
85+
86+ size = MIN (size, (n_bits + GMP_NUMB_BITS - 1) / GMP_NUMB_BITS);
87+
88+ up -= size; /* point at least significant relevant limb */
89+ vp -= size; /* point at least significant relevant limb */
90+
91+ for (i = size - 1; i > 0; i--)
92 {
93 if (up[i] != vp[i])
94 return 0;
95 }
96
97- return 1;
98+ diff = (up[0] ^ vp[0]) >> GMP_NUMB_BITS - 1 - (n_bits - 1) % GMP_NUMB_BITS;
99+ return diff == 0;
100 }
101diff -Naur gmp-4.2.4.orig/mpf/set_str.c gmp-4.2.4/mpf/set_str.c
102--- gmp-4.2.4.orig/mpf/set_str.c 2008-08-25 10:11:37.000000000 -0400
103+++ gmp-4.2.4/mpf/set_str.c 2009-01-03 13:48:18.493274358 -0500
104@@ -137,7 +137,12 @@
105 c = (unsigned char) *++str;
106 }
107
108+ /* Default base to decimal. */
109+ if (base == 0)
110+ base = 10;
111+
112 exp_base = base;
113+
114 if (base < 0)
115 {
116 exp_base = 10;
117@@ -165,10 +170,6 @@
118 return -1;
119 }
120
121- /* Default base to decimal. */
122- if (base == 0)
123- base = 10;
124-
125 /* Locate exponent part of the input. Look from the right of the string,
126 since the exponent is usually a lot shorter than the mantissa. */
127 expptr = NULL;
128diff -Naur gmp-4.2.4.orig/mpz/perfpow.c gmp-4.2.4/mpz/perfpow.c
129--- gmp-4.2.4.orig/mpz/perfpow.c 2007-08-30 14:31:41.000000000 -0400
130+++ gmp-4.2.4/mpz/perfpow.c 2009-01-03 13:47:51.611742467 -0500
131@@ -1,7 +1,7 @@
132 /* mpz_perfect_power_p(arg) -- Return non-zero if ARG is a perfect power,
133 zero otherwise.
134
135-Copyright 1998, 1999, 2000, 2001, 2005 Free Software Foundation, Inc.
136+Copyright 1998, 1999, 2000, 2001, 2005, 2008 Free Software Foundation, Inc.
137
138 This file is part of the GNU MP Library.
139
140@@ -59,6 +59,8 @@
141 #define SMALLEST_OMITTED_PRIME 1009
142
143
144+#define POW2P(a) (((a) & ((a) - 1)) == 0)
145+
146 int
147 mpz_perfect_power_p (mpz_srcptr u)
148 {
149@@ -72,16 +74,13 @@
150 mp_size_t usize = SIZ (u);
151 TMP_DECL;
152
153- if (usize == 0)
154- return 1; /* consider 0 a perfect power */
155+ if (mpz_cmpabs_ui (u, 1) <= 0)
156+ return 1; /* -1, 0, and +1 are perfect powers */
157
158 n2 = mpz_scan1 (u, 0);
159 if (n2 == 1)
160 return 0; /* 2 divides exactly once. */
161
162- if (n2 != 0 && (n2 & 1) == 0 && usize < 0)
163- return 0; /* 2 has even multiplicity with negative U */
164-
165 TMP_MARK;
166
167 uns = ABS (usize) - n2 / BITS_PER_MP_LIMB;
168@@ -89,6 +88,14 @@
169 MPZ_TMP_INIT (u2, uns);
170
171 mpz_tdiv_q_2exp (u2, u, n2);
172+ mpz_abs (u2, u2);
173+
174+ if (mpz_cmp_ui (u2, 1) == 0)
175+ {
176+ TMP_FREE;
177+ /* factoring completed; consistent power */
178+ return ! (usize < 0 && POW2P(n2));
179+ }
180
181 if (isprime (n2))
182 goto n2prime;
183@@ -97,6 +104,9 @@
184 {
185 prime = primes[i];
186
187+ if (mpz_cmp_ui (u2, prime) < 0)
188+ break;
189+
190 if (mpz_divisible_ui_p (u2, prime)) /* divisible by this prime? */
191 {
192 rem = mpz_tdiv_q_ui (q, u2, prime * prime);
193@@ -115,12 +125,6 @@
194 n++;
195 }
196
197- if ((n & 1) == 0 && usize < 0)
198- {
199- TMP_FREE;
200- return 0; /* even multiplicity with negative U, reject */
201- }
202-
203 n2 = gcd (n2, n);
204 if (n2 == 1)
205 {
206@@ -128,10 +132,11 @@
207 return 0; /* we have multiplicity 1 of some factor */
208 }
209
210- if (mpz_cmpabs_ui (u2, 1) == 0)
211+ if (mpz_cmp_ui (u2, 1) == 0)
212 {
213 TMP_FREE;
214- return 1; /* factoring completed; consistent power */
215+ /* factoring completed; consistent power */
216+ return ! (usize < 0 && POW2P(n2));
217 }
218
219 /* As soon as n2 becomes a prime number, stop factoring.
220@@ -169,6 +174,10 @@
221 else
222 {
223 unsigned long int nth;
224+
225+ if (usize < 0 && POW2P(n2))
226+ return 0;
227+
228 /* We found some factors above. We just need to consider values of n
229 that divides n2. */
230 for (nth = 2; nth <= n2; nth++)
231@@ -184,8 +193,11 @@
232 exact = mpz_root (q, u2, nth);
233 if (exact)
234 {
235- TMP_FREE;
236- return 1;
237+ if (! (usize < 0 && POW2P(nth)))
238+ {
239+ TMP_FREE;
240+ return 1;
241+ }
242 }
243 if (mpz_cmp_ui (q, SMALLEST_OMITTED_PRIME) < 0)
244 {
245@@ -199,6 +211,9 @@
246 }
247
248 n2prime:
249+ if (usize < 0 && POW2P(n2))
250+ return 0;
251+
252 exact = mpz_root (NULL, u2, n2);
253 TMP_FREE;
254 return exact;
255diff -Naur gmp-4.2.4.orig/tests/cxx/t-prec.cc gmp-4.2.4/tests/cxx/t-prec.cc
256--- gmp-4.2.4.orig/tests/cxx/t-prec.cc 2007-09-01 06:09:03.000000000 -0400
257+++ gmp-4.2.4/tests/cxx/t-prec.cc 2009-01-03 13:48:39.498471376 -0500
258@@ -1,6 +1,6 @@
259 /* Test precision of mpf_class expressions.
260
261-Copyright 2001, 2002, 2003 Free Software Foundation, Inc.
262+Copyright 2001, 2002, 2003, 2008 Free Software Foundation, Inc.
263
264 This file is part of the GNU MP Library.
265
266@@ -61,7 +61,7 @@
267 g = 1 / f;
268 ASSERT_ALWAYS_PREC
269 (g, "0.11111 11111 11111 11111 11111 11111 11111 11111 11111 11111"
270- " 11111 11111 11111 11111 11111 11", very_large_prec);
271+ " 11111 11111 11111 11111 11111 111", very_large_prec);
272 }
273 {
274 mpf_class f(15.0, large_prec);
275@@ -69,7 +69,7 @@
276 g = 1 / f;
277 ASSERT_ALWAYS_PREC
278 (g, "0.06666 66666 66666 66666 66666 66666 66666 66666 66666 66666"
279- " 66666 66666 66666 66666 66666 67", very_large_prec);
280+ " 66666 66666 66666 66666 66666 667", very_large_prec);
281 }
282
283 // compound expressions
284@@ -94,14 +94,14 @@
285 i = f / g + h;
286 ASSERT_ALWAYS_PREC
287 (i, "15.33333 33333 33333 33333 33333 33333 33333 33333 33333 33333"
288- " 33333 33333 33333 333", very_large_prec);
289+ " 33333 33333 33333 33333 33333 3", very_large_prec);
290 }
291 {
292 mpf_class f(3.0, small_prec);
293 mpf_class g(-(1 + f) / 3, very_large_prec);
294 ASSERT_ALWAYS_PREC
295 (g, "-1.33333 33333 33333 33333 33333 33333 33333 33333 33333 33333"
296- " 33333 33333 33333 333", very_large_prec);
297+ " 33333 33333 33333 33333 33333 33", very_large_prec);
298 }
299 {
300 mpf_class f(9.0, medium_prec);
301@@ -117,7 +117,7 @@
302 g = hypot(1 + 5 / f, 1.0);
303 ASSERT_ALWAYS_PREC
304 (g, "1.66666 66666 66666 66666 66666 66666 66666 66666 66666 66666"
305- " 66666 66666 66666 667", very_large_prec);
306+ " 66666 66666 66666 66666 66666 67", very_large_prec);
307 }
308
309 // compound assignments
310@@ -142,7 +142,7 @@
311 mpf_class g(0.0, very_large_prec);
312 g = mpf_class(1 / f);
313 ASSERT_ALWAYS_PREC
314- (g, "0.11111 11111 11111 11111 11111 11111 11111 111", medium_prec);
315+ (g, "0.11111 11111 11111 11111 11111 11111 11111 1111", medium_prec);
316 }
317 {
318 mpf_class f(15.0, large_prec);
319@@ -150,7 +150,7 @@
320 g = mpf_class(1 / f);
321 ASSERT_ALWAYS_PREC
322 (g, "0.06666 66666 66666 66666 66666 66666 66666 66666 66666 66666"
323- " 66666 667", large_prec);
324+ " 66666 6667", large_prec);
325 }
326
327 {
328@@ -158,7 +158,8 @@
329 mpf_class h(0.0, very_large_prec);
330 h = mpf_class(f / g + 1, large_prec);
331 ASSERT_ALWAYS_PREC
332- (h, "1.33333 33333 33333 33333 33333 33333 33333 33333 33333 3333",
333+ (h, "1.33333 33333 33333 33333 33333 33333 33333 33333 33333 33333"
334+ " 33333 333",
335 large_prec);
336 }
337
338@@ -170,7 +171,7 @@
339 g = f - q;
340 ASSERT_ALWAYS_PREC
341 (g, "2.66666 66666 66666 66666 66666 66666 66666 66666 66666 66666"
342- " 66666 66666 66666 667", very_large_prec);
343+ " 66666 66666 66666 66666 66666 67", very_large_prec);
344 }
345
346 {
347@@ -179,7 +180,8 @@
348 mpf_class g(0.0, very_large_prec);
349 g = mpf_class(f - q, large_prec);
350 ASSERT_ALWAYS_PREC
351- (g, "2.66666 66666 66666 66666 66666 66666 66666 66666 66666 6667",
352+ (g, "2.66666 66666 66666 66666 66666 66666 66666 66666 66666 66666"
353+ " 66666 667",
354 large_prec);
355 }
356 {
357@@ -188,7 +190,7 @@
358 mpf_class g(0.0, very_large_prec);
359 g = mpf_class(f - q);
360 ASSERT_ALWAYS_PREC
361- (g, "2.66666 66666 66666 66666 66666 6667", medium_prec);
362+ (g, "2.66666 66666 66666 66666 66666 66666 66666 667", medium_prec);
363 }
364 {
365 mpf_class f(15.0, large_prec);
366@@ -196,7 +198,8 @@
367 mpf_class g(0.0, very_large_prec);
368 g = mpf_class(f + q);
369 ASSERT_ALWAYS_PREC
370- (g, "15.33333 33333 33333 33333 33333 33333 33333 33333 33333 3333",
371+ (g, "15.33333 33333 33333 33333 33333 33333 33333 33333 33333 33333"
372+ " 33333 33",
373 large_prec);
374 }
375 }
Note: See TracBrowser for help on using the repository browser.