xref: /openbmc/linux/arch/mips/math-emu/dp_maddf.c (revision dc6a81c3)
1 // SPDX-License-Identifier: GPL-2.0-only
2 /*
3  * IEEE754 floating point arithmetic
4  * double precision: MADDF.f (Fused Multiply Add)
5  * MADDF.fmt: FPR[fd] = FPR[fd] + (FPR[fs] x FPR[ft])
6  *
7  * MIPS floating point support
8  * Copyright (C) 2015 Imagination Technologies, Ltd.
9  * Author: Markos Chandras <markos.chandras@imgtec.com>
10  */
11 
12 #include "ieee754dp.h"
13 
14 
15 /* 128 bits shift right logical with rounding. */
16 static void srl128(u64 *hptr, u64 *lptr, int count)
17 {
18 	u64 low;
19 
20 	if (count >= 128) {
21 		*lptr = *hptr != 0 || *lptr != 0;
22 		*hptr = 0;
23 	} else if (count >= 64) {
24 		if (count == 64) {
25 			*lptr = *hptr | (*lptr != 0);
26 		} else {
27 			low = *lptr;
28 			*lptr = *hptr >> (count - 64);
29 			*lptr |= (*hptr << (128 - count)) != 0 || low != 0;
30 		}
31 		*hptr = 0;
32 	} else {
33 		low = *lptr;
34 		*lptr = low >> count | *hptr << (64 - count);
35 		*lptr |= (low << (64 - count)) != 0;
36 		*hptr = *hptr >> count;
37 	}
38 }
39 
40 static union ieee754dp _dp_maddf(union ieee754dp z, union ieee754dp x,
41 				 union ieee754dp y, enum maddf_flags flags)
42 {
43 	int re;
44 	int rs;
45 	unsigned int lxm;
46 	unsigned int hxm;
47 	unsigned int lym;
48 	unsigned int hym;
49 	u64 lrm;
50 	u64 hrm;
51 	u64 lzm;
52 	u64 hzm;
53 	u64 t;
54 	u64 at;
55 	int s;
56 
57 	COMPXDP;
58 	COMPYDP;
59 	COMPZDP;
60 
61 	EXPLODEXDP;
62 	EXPLODEYDP;
63 	EXPLODEZDP;
64 
65 	FLUSHXDP;
66 	FLUSHYDP;
67 	FLUSHZDP;
68 
69 	ieee754_clearcx();
70 
71 	rs = xs ^ ys;
72 	if (flags & MADDF_NEGATE_PRODUCT)
73 		rs ^= 1;
74 	if (flags & MADDF_NEGATE_ADDITION)
75 		zs ^= 1;
76 
77 	/*
78 	 * Handle the cases when at least one of x, y or z is a NaN.
79 	 * Order of precedence is sNaN, qNaN and z, x, y.
80 	 */
81 	if (zc == IEEE754_CLASS_SNAN)
82 		return ieee754dp_nanxcpt(z);
83 	if (xc == IEEE754_CLASS_SNAN)
84 		return ieee754dp_nanxcpt(x);
85 	if (yc == IEEE754_CLASS_SNAN)
86 		return ieee754dp_nanxcpt(y);
87 	if (zc == IEEE754_CLASS_QNAN)
88 		return z;
89 	if (xc == IEEE754_CLASS_QNAN)
90 		return x;
91 	if (yc == IEEE754_CLASS_QNAN)
92 		return y;
93 
94 	if (zc == IEEE754_CLASS_DNORM)
95 		DPDNORMZ;
96 	/* ZERO z cases are handled separately below */
97 
98 	switch (CLPAIR(xc, yc)) {
99 
100 	/*
101 	 * Infinity handling
102 	 */
103 	case CLPAIR(IEEE754_CLASS_INF, IEEE754_CLASS_ZERO):
104 	case CLPAIR(IEEE754_CLASS_ZERO, IEEE754_CLASS_INF):
105 		ieee754_setcx(IEEE754_INVALID_OPERATION);
106 		return ieee754dp_indef();
107 
108 	case CLPAIR(IEEE754_CLASS_NORM, IEEE754_CLASS_INF):
109 	case CLPAIR(IEEE754_CLASS_DNORM, IEEE754_CLASS_INF):
110 	case CLPAIR(IEEE754_CLASS_INF, IEEE754_CLASS_NORM):
111 	case CLPAIR(IEEE754_CLASS_INF, IEEE754_CLASS_DNORM):
112 	case CLPAIR(IEEE754_CLASS_INF, IEEE754_CLASS_INF):
113 		if ((zc == IEEE754_CLASS_INF) && (zs != rs)) {
114 			/*
115 			 * Cases of addition of infinities with opposite signs
116 			 * or subtraction of infinities with same signs.
117 			 */
118 			ieee754_setcx(IEEE754_INVALID_OPERATION);
119 			return ieee754dp_indef();
120 		}
121 		/*
122 		 * z is here either not an infinity, or an infinity having the
123 		 * same sign as product (x*y). The result must be an infinity,
124 		 * and its sign is determined only by the sign of product (x*y).
125 		 */
126 		return ieee754dp_inf(rs);
127 
128 	case CLPAIR(IEEE754_CLASS_ZERO, IEEE754_CLASS_ZERO):
129 	case CLPAIR(IEEE754_CLASS_ZERO, IEEE754_CLASS_NORM):
130 	case CLPAIR(IEEE754_CLASS_ZERO, IEEE754_CLASS_DNORM):
131 	case CLPAIR(IEEE754_CLASS_NORM, IEEE754_CLASS_ZERO):
132 	case CLPAIR(IEEE754_CLASS_DNORM, IEEE754_CLASS_ZERO):
133 		if (zc == IEEE754_CLASS_INF)
134 			return ieee754dp_inf(zs);
135 		if (zc == IEEE754_CLASS_ZERO) {
136 			/* Handle cases +0 + (-0) and similar ones. */
137 			if (zs == rs)
138 				/*
139 				 * Cases of addition of zeros of equal signs
140 				 * or subtraction of zeroes of opposite signs.
141 				 * The sign of the resulting zero is in any
142 				 * such case determined only by the sign of z.
143 				 */
144 				return z;
145 
146 			return ieee754dp_zero(ieee754_csr.rm == FPU_CSR_RD);
147 		}
148 		/* x*y is here 0, and z is not 0, so just return z */
149 		return z;
150 
151 	case CLPAIR(IEEE754_CLASS_DNORM, IEEE754_CLASS_DNORM):
152 		DPDNORMX;
153 		/* fall through */
154 
155 	case CLPAIR(IEEE754_CLASS_NORM, IEEE754_CLASS_DNORM):
156 		if (zc == IEEE754_CLASS_INF)
157 			return ieee754dp_inf(zs);
158 		DPDNORMY;
159 		break;
160 
161 	case CLPAIR(IEEE754_CLASS_DNORM, IEEE754_CLASS_NORM):
162 		if (zc == IEEE754_CLASS_INF)
163 			return ieee754dp_inf(zs);
164 		DPDNORMX;
165 		break;
166 
167 	case CLPAIR(IEEE754_CLASS_NORM, IEEE754_CLASS_NORM):
168 		if (zc == IEEE754_CLASS_INF)
169 			return ieee754dp_inf(zs);
170 		/* continue to real computations */
171 	}
172 
173 	/* Finally get to do some computation */
174 
175 	/*
176 	 * Do the multiplication bit first
177 	 *
178 	 * rm = xm * ym, re = xe + ye basically
179 	 *
180 	 * At this point xm and ym should have been normalized.
181 	 */
182 	assert(xm & DP_HIDDEN_BIT);
183 	assert(ym & DP_HIDDEN_BIT);
184 
185 	re = xe + ye;
186 
187 	/* shunt to top of word */
188 	xm <<= 64 - (DP_FBITS + 1);
189 	ym <<= 64 - (DP_FBITS + 1);
190 
191 	/*
192 	 * Multiply 64 bits xm and ym to give 128 bits result in hrm:lrm.
193 	 */
194 
195 	lxm = xm;
196 	hxm = xm >> 32;
197 	lym = ym;
198 	hym = ym >> 32;
199 
200 	lrm = DPXMULT(lxm, lym);
201 	hrm = DPXMULT(hxm, hym);
202 
203 	t = DPXMULT(lxm, hym);
204 
205 	at = lrm + (t << 32);
206 	hrm += at < lrm;
207 	lrm = at;
208 
209 	hrm = hrm + (t >> 32);
210 
211 	t = DPXMULT(hxm, lym);
212 
213 	at = lrm + (t << 32);
214 	hrm += at < lrm;
215 	lrm = at;
216 
217 	hrm = hrm + (t >> 32);
218 
219 	/* Put explicit bit at bit 126 if necessary */
220 	if ((int64_t)hrm < 0) {
221 		lrm = (hrm << 63) | (lrm >> 1);
222 		hrm = hrm >> 1;
223 		re++;
224 	}
225 
226 	assert(hrm & (1 << 62));
227 
228 	if (zc == IEEE754_CLASS_ZERO) {
229 		/*
230 		 * Move explicit bit from bit 126 to bit 55 since the
231 		 * ieee754dp_format code expects the mantissa to be
232 		 * 56 bits wide (53 + 3 rounding bits).
233 		 */
234 		srl128(&hrm, &lrm, (126 - 55));
235 		return ieee754dp_format(rs, re, lrm);
236 	}
237 
238 	/* Move explicit bit from bit 52 to bit 126 */
239 	lzm = 0;
240 	hzm = zm << 10;
241 	assert(hzm & (1 << 62));
242 
243 	/* Make the exponents the same */
244 	if (ze > re) {
245 		/*
246 		 * Have to shift y fraction right to align.
247 		 */
248 		s = ze - re;
249 		srl128(&hrm, &lrm, s);
250 		re += s;
251 	} else if (re > ze) {
252 		/*
253 		 * Have to shift x fraction right to align.
254 		 */
255 		s = re - ze;
256 		srl128(&hzm, &lzm, s);
257 		ze += s;
258 	}
259 	assert(ze == re);
260 	assert(ze <= DP_EMAX);
261 
262 	/* Do the addition */
263 	if (zs == rs) {
264 		/*
265 		 * Generate 128 bit result by adding two 127 bit numbers
266 		 * leaving result in hzm:lzm, zs and ze.
267 		 */
268 		hzm = hzm + hrm + (lzm > (lzm + lrm));
269 		lzm = lzm + lrm;
270 		if ((int64_t)hzm < 0) {        /* carry out */
271 			srl128(&hzm, &lzm, 1);
272 			ze++;
273 		}
274 	} else {
275 		if (hzm > hrm || (hzm == hrm && lzm >= lrm)) {
276 			hzm = hzm - hrm - (lzm < lrm);
277 			lzm = lzm - lrm;
278 		} else {
279 			hzm = hrm - hzm - (lrm < lzm);
280 			lzm = lrm - lzm;
281 			zs = rs;
282 		}
283 		if (lzm == 0 && hzm == 0)
284 			return ieee754dp_zero(ieee754_csr.rm == FPU_CSR_RD);
285 
286 		/*
287 		 * Put explicit bit at bit 126 if necessary.
288 		 */
289 		if (hzm == 0) {
290 			/* left shift by 63 or 64 bits */
291 			if ((int64_t)lzm < 0) {
292 				/* MSB of lzm is the explicit bit */
293 				hzm = lzm >> 1;
294 				lzm = lzm << 63;
295 				ze -= 63;
296 			} else {
297 				hzm = lzm;
298 				lzm = 0;
299 				ze -= 64;
300 			}
301 		}
302 
303 		t = 0;
304 		while ((hzm >> (62 - t)) == 0)
305 			t++;
306 
307 		assert(t <= 62);
308 		if (t) {
309 			hzm = hzm << t | lzm >> (64 - t);
310 			lzm = lzm << t;
311 			ze -= t;
312 		}
313 	}
314 
315 	/*
316 	 * Move explicit bit from bit 126 to bit 55 since the
317 	 * ieee754dp_format code expects the mantissa to be
318 	 * 56 bits wide (53 + 3 rounding bits).
319 	 */
320 	srl128(&hzm, &lzm, (126 - 55));
321 
322 	return ieee754dp_format(zs, ze, lzm);
323 }
324 
325 union ieee754dp ieee754dp_maddf(union ieee754dp z, union ieee754dp x,
326 				union ieee754dp y)
327 {
328 	return _dp_maddf(z, x, y, 0);
329 }
330 
331 union ieee754dp ieee754dp_msubf(union ieee754dp z, union ieee754dp x,
332 				union ieee754dp y)
333 {
334 	return _dp_maddf(z, x, y, MADDF_NEGATE_PRODUCT);
335 }
336 
337 union ieee754dp ieee754dp_madd(union ieee754dp z, union ieee754dp x,
338 				union ieee754dp y)
339 {
340 	return _dp_maddf(z, x, y, 0);
341 }
342 
343 union ieee754dp ieee754dp_msub(union ieee754dp z, union ieee754dp x,
344 				union ieee754dp y)
345 {
346 	return _dp_maddf(z, x, y, MADDF_NEGATE_ADDITION);
347 }
348 
349 union ieee754dp ieee754dp_nmadd(union ieee754dp z, union ieee754dp x,
350 				union ieee754dp y)
351 {
352 	return _dp_maddf(z, x, y, MADDF_NEGATE_PRODUCT|MADDF_NEGATE_ADDITION);
353 }
354 
355 union ieee754dp ieee754dp_nmsub(union ieee754dp z, union ieee754dp x,
356 				union ieee754dp y)
357 {
358 	return _dp_maddf(z, x, y, MADDF_NEGATE_PRODUCT);
359 }
360