middle-end/7151: ICE when compiling for UltraSPARC
gawrilow@math.TU-Berlin.DE
gawrilow@math.TU-Berlin.DE
Fri Jun 28 03:56:00 GMT 2002
>Number: 7151
>Category: middle-end
>Synopsis: ICE when compiling for UltraSPARC
>Confidential: no
>Severity: serious
>Priority: medium
>Responsible: unassigned
>State: open
>Class: ice-on-legal-code
>Submitter-Id: net
>Arrival-Date: Fri Jun 28 02:16:00 PDT 2002
>Closed-Date:
>Last-Modified:
>Originator: Ewgenij Gawrilow
>Release: 3.1
>Organization:
TU Berlin
>Environment:
System: SunOS jabba 5.8 Generic_108528-13 sun4u sparc SUNW,UltraAX-MP
Architecture: sun4
host: sparc-sun-solaris2.8
build: sparc-sun-solaris2.8
target: sparc-sun-solaris2.8
configured with: /homes/combi/gawrilow/tmp/gcc-3.1/configure --prefix=/usr/site-local/experimental --with-local-prefix=/usr/site-local --with-gxx-include-dir=/usr/site-local/share/include/g++ --enable-shared --enable-libgcj --enable-languages=c++,java --with-cpu=ultrasparc
>Description:
I have tried to compile Mozilla 1.0 from the sources for the
UltraSPARC platform. It got stuck at the following file:
gcc -o e_pow.o -c -DOSTYPE=\"SunOS5\" -DOSARCH=\"SunOS\" -DOJI -I../../../dist/include/js -I../../../dist/include -I/net/jabba.opt/combi/gawrilow/mozilla/dist/include/nspr -I/usr/lib/include -I/usr/openwin/include -fPIC -I/usr/openwin/include -pedantic -Wno-long-long -mcpu=ultrasparc -fshort-wchar -pthreads -DNDEBUG -DTRIMMED -O3 -D_IEEE_LIBM -I/usr/openwin/include -include ../../../config-defs.h -DMOZILLA_CLIENT -Wp,-MD,.deps/e_pow.pp e_pow.c
e_pow.c: In function `__ieee754_pow':
e_pow.c:376: insn does not satisfy its constraints:
(insn 1875 777 1878 (set (reg:DI 34 %f2)
(reg/v:DI 64 %f32 [112])) 58 {*movdi_insn_sp32_v9} (nil)
(nil))
e_pow.c:376: Internal compiler error in reload_cse_simplify_operands, at reload1.c:8368
>How-To-Repeat:
# 1 "e_pow.c"
# 1 "<built-in>"
# 1 "<command line>"
# 1 "../../../config-defs.h" 1
/* List of defines generated by configure. Included with preprocessor flag,
* -include, to avoid long list of -D defines on the compile command-line.
* Do not edit.
*/
# 2 "<command line>" 2
# 1 "e_pow.c"
/* -*- Mode: C; tab-width: 8; indent-tabs-mode: nil; c-basic-offset: 4 -*-
*
* The contents of this file are subject to the Netscape Public
* License Version 1.1 (the "License"); you may not use this file
* except in compliance with the License. You may obtain a copy of
* the License at http://www.mozilla.org/NPL/
*
* Software distributed under the License is distributed on an "AS
* IS" basis, WITHOUT WARRANTY OF ANY KIND, either express or
* implied. See the License for the specific language governing
* rights and limitations under the License.
*
* The Original Code is Mozilla Communicator client code, released
* March 31, 1998.
*
* The Initial Developer of the Original Code is Sun Microsystems,
* Inc. Portions created by Netscape are
* Copyright (C) 1998 Netscape Communications Corporation. All
* Rights Reserved.
*
* Contributor(s):
*
* Alternatively, the contents of this file may be used under the
* terms of the GNU Public License (the "GPL"), in which case the
* provisions of the GPL are applicable instead of those above.
* If you wish to allow use of your version of this file only
* under the terms of the GPL and not to allow others to use your
* version of this file under the NPL, indicate your decision by
* deleting the provisions above and replace them with the notice
* and other provisions required by the GPL. If you do not delete
* the provisions above, a recipient may use your version of this
* file under either the NPL or the GPL.
*/
/* @(#)e_pow.c 1.3 95/01/18 */
/*
* ====================================================
* Copyright (C) 1993 by Sun Microsystems, Inc. All rights reserved.
*
* Developed at SunSoft, a Sun Microsystems, Inc. business.
* Permission to use, copy, modify, and distribute this
* software is freely granted, provided that this notice
* is preserved.
* ====================================================
*/
/* __ieee754_pow(x,y) return x**y
*
* n
* Method: Let x = 2 * (1+f)
* 1. Compute and return log2(x) in two pieces:
* log2(x) = w1 + w2,
* where w1 has 53-24 = 29 bit trailing zeros.
* 2. Perform y*log2(x) = n+y' by simulating muti-precision
* arithmetic, where |y'|<=0.5.
* 3. Return x**y = 2**n*exp(y'*log2)
*
* Special cases:
* 1. (anything) ** 0 is 1
* 2. (anything) ** 1 is itself
* 3. (anything) ** NAN is NAN
* 4. NAN ** (anything except 0) is NAN
* 5. +-(|x| > 1) ** +INF is +INF
* 6. +-(|x| > 1) ** -INF is +0
* 7. +-(|x| < 1) ** +INF is +0
* 8. +-(|x| < 1) ** -INF is +INF
* 9. +-1 ** +-INF is NAN
* 10. +0 ** (+anything except 0, NAN) is +0
* 11. -0 ** (+anything except 0, NAN, odd integer) is +0
* 12. +0 ** (-anything except 0, NAN) is +INF
* 13. -0 ** (-anything except 0, NAN, odd integer) is +INF
* 14. -0 ** (odd integer) = -( +0 ** (odd integer) )
* 15. +INF ** (+anything except 0,NAN) is +INF
* 16. +INF ** (-anything except 0,NAN) is +0
* 17. -INF ** (anything) = -0 ** (-anything)
* 18. (-anything) ** (integer) is (-1)**(integer)*(+anything**integer)
* 19. (-anything except 0 and inf) ** (non-integer) is NAN
*
* Accuracy:
* pow(x,y) returns x**y nearly rounded. In particular
* pow(integer,integer)
* always returns the correct integer provided it is
* representable.
*
* Constants :
* The hexadecimal values are the intended ones for the following
* constants. The decimal values may be used, provided that the
* compiler will convert from decimal to binary accurately enough
* to produce the hexadecimal values shown.
*/
# 1 "fdlibm.h" 1
/* -*- Mode: C; tab-width: 8; indent-tabs-mode: nil; c-basic-offset: 4 -*-
*
* The contents of this file are subject to the Netscape Public
* License Version 1.1 (the "License"); you may not use this file
* except in compliance with the License. You may obtain a copy of
* the License at http://www.mozilla.org/NPL/
*
* Software distributed under the License is distributed on an "AS
* IS" basis, WITHOUT WARRANTY OF ANY KIND, either express or
* implied. See the License for the specific language governing
* rights and limitations under the License.
*
* The Original Code is Mozilla Communicator client code, released
* March 31, 1998.
*
* The Initial Developer of the Original Code is Sun Microsystems,
* Inc. Portions created by Netscape are
* Copyright (C) 1998 Netscape Communications Corporation. All
* Rights Reserved.
*
* Contributor(s):
*
* Alternatively, the contents of this file may be used under the
* terms of the GNU Public License (the "GPL"), in which case the
* provisions of the GPL are applicable instead of those above.
* If you wish to allow use of your version of this file only
* under the terms of the GPL and not to allow others to use your
* version of this file under the NPL, indicate your decision by
* deleting the provisions above and replace them with the notice
* and other provisions required by the GPL. If you do not delete
* the provisions above, a recipient may use your version of this
* file under either the NPL or the GPL.
*/
/* @(#)fdlibm.h 1.5 95/01/18 */
/*
* ====================================================
* Copyright (C) 1993 by Sun Microsystems, Inc. All rights reserved.
*
* Developed at SunSoft, a Sun Microsystems, Inc. business.
* Permission to use, copy, modify, and distribute this
* software is freely granted, provided that this notice
* is preserved.
* ====================================================
*/
/* Modified defines start here.. */
# 63 "fdlibm.h"
/* End here. The rest is the standard file. */
# 72 "fdlibm.h"
typedef union {
struct { int hi, lo; } ints;
double d;
} fd_twoints;
# 91 "fdlibm.h"
/*
* ANSI/POSIX
*/
extern int signgam;
enum fdversion {fdlibm_ieee = -1, fdlibm_svid, fdlibm_xopen, fdlibm_posix};
/* if global variable _LIB_VERSION is not desirable, one may
* change the following to be a constant by:
* #define _LIB_VERSION_TYPE const enum version
* In that case, after one initializes the value _LIB_VERSION (see
* s_lib_version.c) during compile time, it cannot be modified
* in the middle of a program
*/
extern enum fdversion _fdlib_version;
struct exception {
int type;
char *name;
double arg1;
double arg2;
double retval;
};
/*
* set X_TLOSS = pi*2**52, which is possibly defined in <values.h>
* (one may replace the following line by "#include <values.h>")
*/
# 142 "fdlibm.h"
/*
* ANSI/POSIX
*/
extern double fd_acos (double);
extern double fd_asin (double);
extern double fd_atan (double);
extern double fd_atan2 (double, double);
extern double fd_cos (double);
extern double fd_sin (double);
extern double fd_tan (double);
extern double fd_cosh (double);
extern double fd_sinh (double);
extern double fd_tanh (double);
extern double fd_exp (double);
extern double fd_frexp (double, int *);
extern double fd_ldexp (double, int);
extern double fd_log (double);
extern double fd_log10 (double);
extern double fd_modf (double, double *);
extern double fd_pow (double, double);
extern double fd_sqrt (double);
extern double fd_ceil (double);
extern double fd_fabs (double);
extern double fd_floor (double);
extern double fd_fmod (double, double);
extern double fd_erf (double);
extern double fd_erfc (double);
extern double fd_gamma (double);
extern double fd_hypot (double, double);
extern int fd_isnan (double);
extern int fd_finite (double);
extern double fd_j0 (double);
extern double fd_j1 (double);
extern double fd_jn (int, double);
extern double fd_lgamma (double);
extern double fd_y0 (double);
extern double fd_y1 (double);
extern double fd_yn (int, double);
extern double fd_acosh (double);
extern double fd_asinh (double);
extern double fd_atanh (double);
extern double fd_cbrt (double);
extern double fd_logb (double);
extern double fd_nextafter (double, double);
extern double fd_remainder (double, double);
extern double fd_scalb (double, double);
extern int fd_matherr (struct exception *);
/*
* IEEE Test Vector
*/
extern double significand (double);
/*
* Functions callable from C, intended to support IEEE arithmetic.
*/
extern double fd_copysign (double, double);
extern int fd_ilogb (double);
extern double fd_rint (double);
extern double fd_scalbn (double, int);
/*
* BSD math library entry points
*/
extern double fd_expm1 (double);
extern double fd_log1p (double);
/*
* Reentrant version of gamma & lgamma; passes signgam back by reference
* as the second argument; user must allocate space for signgam.
*/
extern double gamma_r (double, int *);
extern double lgamma_r (double, int *);
/* ieee style elementary functions */
extern double __ieee754_sqrt (double);
extern double __ieee754_acos (double);
extern double __ieee754_acosh (double);
extern double __ieee754_log (double);
extern double __ieee754_atanh (double);
extern double __ieee754_asin (double);
extern double __ieee754_atan2 (double,double);
extern double __ieee754_exp (double);
extern double __ieee754_cosh (double);
extern double __ieee754_fmod (double,double);
extern double __ieee754_pow (double,double);
extern double __ieee754_lgamma_r (double,int *);
extern double __ieee754_gamma_r (double,int *);
extern double __ieee754_lgamma (double);
extern double __ieee754_gamma (double);
extern double __ieee754_log10 (double);
extern double __ieee754_sinh (double);
extern double __ieee754_hypot (double,double);
extern double __ieee754_j0 (double);
extern double __ieee754_j1 (double);
extern double __ieee754_y0 (double);
extern double __ieee754_y1 (double);
extern double __ieee754_jn (int,double);
extern double __ieee754_yn (int,double);
extern double __ieee754_remainder (double,double);
extern int __ieee754_rem_pio2 (double,double*);
extern double __ieee754_scalb (double,double);
/* fdlibm kernel function */
extern double __kernel_standard (double,double,int,int*);
extern double __kernel_sin (double,double,int);
extern double __kernel_cos (double,double);
extern double __kernel_tan (double,double,int);
extern int __kernel_rem_pio2 (double*,double*,int,int,int,const int*);
# 93 "e_pow.c" 2
static const double
bp[] = {1.0, 1.5,},
dp_h[] = { 0.0, 5.84962487220764160156e-01,}, /* 0x3FE2B803, 0x40000000 */
dp_l[] = { 0.0, 1.35003920212974897128e-08,}, /* 0x3E4CFDEB, 0x43CFD006 */
zero = 0.0,
one = 1.0,
two = 2.0,
two53 = 9007199254740992.0, /* 0x43400000, 0x00000000 */
really_big = 1.0e300,
tiny = 1.0e-300,
/* poly coefs for (3/2)*(log(x)-2s-2/3*s**3 */
L1 = 5.99999999999994648725e-01, /* 0x3FE33333, 0x33333303 */
L2 = 4.28571428578550184252e-01, /* 0x3FDB6DB6, 0xDB6FABFF */
L3 = 3.33333329818377432918e-01, /* 0x3FD55555, 0x518F264D */
L4 = 2.72728123808534006489e-01, /* 0x3FD17460, 0xA91D4101 */
L5 = 2.30660745775561754067e-01, /* 0x3FCD864A, 0x93C9DB65 */
L6 = 2.06975017800338417784e-01, /* 0x3FCA7E28, 0x4A454EEF */
P1 = 1.66666666666666019037e-01, /* 0x3FC55555, 0x5555553E */
P2 = -2.77777777770155933842e-03, /* 0xBF66C16C, 0x16BEBD93 */
P3 = 6.61375632143793436117e-05, /* 0x3F11566A, 0xAF25DE2C */
P4 = -1.65339022054652515390e-06, /* 0xBEBBBD41, 0xC5D26BF1 */
P5 = 4.13813679705723846039e-08, /* 0x3E663769, 0x72BEA4D0 */
lg2 = 6.93147180559945286227e-01, /* 0x3FE62E42, 0xFEFA39EF */
lg2_h = 6.93147182464599609375e-01, /* 0x3FE62E43, 0x00000000 */
lg2_l = -1.90465429995776804525e-09, /* 0xBE205C61, 0x0CA86C39 */
ovt = 8.0085662595372944372e-0017, /* -(1024-log2(ovfl+.5ulp)) */
cp = 9.61796693925975554329e-01, /* 0x3FEEC709, 0xDC3A03FD =2/(3ln2) */
cp_h = 9.61796700954437255859e-01, /* 0x3FEEC709, 0xE0000000 =(float)cp */
cp_l = -7.02846165095275826516e-09, /* 0xBE3E2FE0, 0x145B01F5 =tail of cp_h*/
ivln2 = 1.44269504088896338700e+00, /* 0x3FF71547, 0x652B82FE =1/ln2 */
ivln2_h = 1.44269502162933349609e+00, /* 0x3FF71547, 0x60000000 =24b 1/ln2*/
ivln2_l = 1.92596299112661746887e-08; /* 0x3E54AE0B, 0xF85DDF44 =1/ln2 tail*/
double __ieee754_pow(double x, double y)
{
fd_twoints ux, uy, uz;
double y1,t1,p_h,t,z,ax;
double z_h,z_l,p_l;
double t2,r,s,u,v,w;
int i,j,k,yisint,n;
int hx,hy,ix,iy;
unsigned lx,ly;
ux.d = x; uy.d = y;
hx = ux.ints.hi; lx = ux.ints.lo;
hy = uy.ints.hi; ly = uy.ints.lo;
ix = hx&0x7fffffff; iy = hy&0x7fffffff;
/* y==zero: x**0 = 1 */
if((iy|ly)==0) return one;
/* +-NaN return x+y */
if(ix > 0x7ff00000 || ((ix==0x7ff00000)&&(lx!=0)) ||
iy > 0x7ff00000 || ((iy==0x7ff00000)&&(ly!=0)))
return x+y;
/* determine if y is an odd int when x < 0
* yisint = 0 ... y is not an integer
* yisint = 1 ... y is an odd int
* yisint = 2 ... y is an even int
*/
yisint = 0;
if(hx<0) {
if(iy>=0x43400000) yisint = 2; /* even integer y */
else if(iy>=0x3ff00000) {
k = (iy>>20)-0x3ff; /* exponent */
if(k>20) {
j = ly>>(52-k);
if((j<<(52-k))==(int)ly) yisint = 2-(j&1);
} else if(ly==0) {
j = iy>>(20-k);
if((j<<(20-k))==iy) yisint = 2-(j&1);
}
}
}
/* special value of y */
if(ly==0) {
if (iy==0x7ff00000) { /* y is +-inf */
if(((ix-0x3ff00000)|lx)==0)
return y - y; /* inf**+-1 is NaN */
else if (ix >= 0x3ff00000)/* (|x|>1)**+-inf = inf,0 */
return (hy>=0)? y: zero;
else /* (|x|<1)**-,+inf = inf,0 */
return (hy<0)?-y: zero;
}
if(iy==0x3ff00000) { /* y is +-1 */
if(hy<0) return one/x; else return x;
}
if(hy==0x40000000) return x*x; /* y is 2 */
if(hy==0x3fe00000) { /* y is 0.5 */
if(hx>=0) /* x >= +0 */
return fd_sqrt(x);
}
}
ax = fd_fabs(x);
/* special value of x */
if(lx==0) {
if(ix==0x7ff00000||ix==0||ix==0x3ff00000){
z = ax; /*x is +-0,+-inf,+-1*/
if(hy<0) z = one/z; /* z = (1/|x|) */
if(hx<0) {
if(((ix-0x3ff00000)|yisint)==0) {
z = (z-z)/(z-z); /* (-1)**non-int is NaN */
} else if(yisint==1) {
z = -z; /* (x<0)**odd = -(|x|**odd) */
}
}
return z;
}
}
/* (x<0)**(non-int) is NaN */
if((((hx>>31)+1)|yisint)==0) return (x-x)/(x-x);
/* |y| is really_big */
if(iy>0x41e00000) { /* if |y| > 2**31 */
if(iy>0x43f00000){ /* if |y| > 2**64, must o/uflow */
if(ix<=0x3fefffff) return (hy<0)? really_big*really_big:tiny*tiny;
if(ix>=0x3ff00000) return (hy>0)? really_big*really_big:tiny*tiny;
}
/* over/underflow if x is not close to one */
if(ix<0x3fefffff) return (hy<0)? really_big*really_big:tiny*tiny;
if(ix>0x3ff00000) return (hy>0)? really_big*really_big:tiny*tiny;
/* now |1-x| is tiny <= 2**-20, suffice to compute
log(x) by x-x^2/2+x^3/3-x^4/4 */
t = x-1; /* t has 20 trailing zeros */
w = (t*t)*(0.5-t*(0.3333333333333333333333-t*0.25));
u = ivln2_h*t; /* ivln2_h has 21 sig. bits */
v = t*ivln2_l-w*ivln2;
t1 = u+v;
uz.d = t1;
uz.ints.lo = 0;
t1 = uz.d;
t2 = v-(t1-u);
} else {
double s_h,t_h;
double s2,s_l,t_l;
n = 0;
/* take care subnormal number */
if(ix<0x00100000)
{ax *= two53; n -= 53; uz.d = ax; ix = uz.ints.hi; }
n += ((ix)>>20)-0x3ff;
j = ix&0x000fffff;
/* determine interval */
ix = j|0x3ff00000; /* normalize ix */
if(j<=0x3988E) k=0; /* |x|<sqrt(3/2) */
else if(j<0xBB67A) k=1; /* |x|<sqrt(3) */
else {k=0;n+=1;ix -= 0x00100000;}
uz.d = ax;
uz.ints.hi = ix;
ax = uz.d;
/* compute s = s_h+s_l = (x-1)/(x+1) or (x-1.5)/(x+1.5) */
u = ax-bp[k]; /* bp[0]=1.0, bp[1]=1.5 */
v = one/(ax+bp[k]);
s = u*v;
s_h = s;
uz.d = s_h;
uz.ints.lo = 0;
s_h = uz.d;
/* t_h=ax+bp[k] High */
t_h = zero;
uz.d = t_h;
uz.ints.hi=((ix>>1)|0x20000000)+0x00080000+(k<<18);
t_h = uz.d;
t_l = ax - (t_h-bp[k]);
s_l = v*((u-s_h*t_h)-s_h*t_l);
/* compute log(ax) */
s2 = s*s;
r = s2*s2*(L1+s2*(L2+s2*(L3+s2*(L4+s2*(L5+s2*L6)))));
r += s_l*(s_h+s);
s2 = s_h*s_h;
t_h = 3.0+s2+r;
uz.d = t_h;
uz.ints.lo = 0;
t_h = uz.d;
t_l = r-((t_h-3.0)-s2);
/* u+v = s*(1+...) */
u = s_h*t_h;
v = s_l*t_h+t_l*s;
/* 2/(3log2)*(s+...) */
p_h = u+v;
uz.d = p_h;
uz.ints.lo = 0;
p_h = uz.d;
p_l = v-(p_h-u);
z_h = cp_h*p_h; /* cp_h+cp_l = 2/(3*log2) */
z_l = cp_l*p_h+p_l*cp+dp_l[k];
/* log2(ax) = (s+..)*2/(3*log2) = n + dp_h + z_h + z_l */
t = (double)n;
t1 = (((z_h+z_l)+dp_h[k])+t);
uz.d = t1;
uz.ints.lo = 0;
t1 = uz.d;
t2 = z_l-(((t1-t)-dp_h[k])-z_h);
}
s = one; /* s (sign of result -ve**odd) = -1 else = 1 */
if((((hx>>31)+1)|(yisint-1))==0) s = -one;/* (-ve)**(odd int) */
/* split up y into y1+y2 and compute (y1+y2)*(t1+t2) */
y1 = y;
uy.d = y1;
uy.ints.lo = 0;
y1 = uy.d;
p_l = (y-y1)*t1+y*t2;
p_h = y1*t1;
z = p_l+p_h;
uz.d = z;
j = uz.ints.hi;
i = uz.ints.lo;
if (j>=0x40900000) { /* z >= 1024 */
if(((j-0x40900000)|i)!=0) /* if z > 1024 */
return s*really_big*really_big; /* overflow */
else {
if(p_l+ovt>z-p_h) return s*really_big*really_big; /* overflow */
}
} else if((j&0x7fffffff)>=0x4090cc00 ) { /* z <= -1075 */
if(((j-0xc090cc00)|i)!=0) /* z < -1075 */
return s*tiny*tiny; /* underflow */
else {
if(p_l<=z-p_h) return s*tiny*tiny; /* underflow */
}
}
/*
* compute 2**(p_h+p_l)
*/
i = j&0x7fffffff;
k = (i>>20)-0x3ff;
n = 0;
if(i>0x3fe00000) { /* if |z| > 0.5, set n = [z+0.5] */
n = j+(0x00100000>>(k+1));
k = ((n&0x7fffffff)>>20)-0x3ff; /* new k for n */
t = zero;
uz.d = t;
uz.ints.hi = (n&~(0x000fffff>>k));
t = uz.d;
n = ((n&0x000fffff)|0x00100000)>>(20-k);
if(j<0) n = -n;
p_h -= t;
}
t = p_l+p_h;
uz.d = t;
uz.ints.lo = 0;
t = uz.d;
u = t*lg2_h;
v = (p_l-(t-p_h))*lg2+t*lg2_l;
z = u+v;
w = v-(z-u);
t = z*z;
t1 = z - t*(P1+t*(P2+t*(P3+t*(P4+t*P5))));
r = (z*t1)/(t1-two)-(w+z*w);
z = one-(r-z);
uz.d = z;
j = uz.ints.hi;
j += (n<<20);
if((j>>20)<=0) z = fd_scalbn(z,n); /* subnormal output */
else { uz.d = z; uz.ints.hi += (n<<20); z = uz.d; }
return s*z;
}
>Fix:
Compilation with -mcpu=supersparc succeeded.
>Release-Note:
>Audit-Trail:
>Unformatted:
More information about the Gcc-bugs
mailing list