JDK-5033578 : Java should require use of latest fdlibm 5.3
  • Type: Bug
  • Component: core-libs
  • Sub-Component: java.lang
  • Affected Version: 1.4.2,5.0
  • Priority: P3
  • Status: Resolved
  • Resolution: Fixed
  • OS: generic,windows_2000
  • CPU: generic,x86
  • Submitted: 2004-04-19
  • Updated: 2017-05-16
  • Resolved: 2004-05-11
The Version table provides details related to the release that this issue/RFE will be addressed.

Unresolved : Release in which this issue/RFE will be addressed.
Resolved: Release in which this issue/RFE has been resolved.
Fixed : Release in which this issue/RFE has been fixed. The release containing this fix may be available for download as an Early Access Release or a General Availability Release.

To download the current JDK release, click here.
Other
5.0 b51Fixed
Related Reports
Duplicate :  
Description
For many years, Java has required use of the algorithms from fdlibm 5.2.  However, fdlibm 5.2 has several bugs in corner cases of pow and tan.  Now that a corrected version of fdlibm is available, the StrictMath spec should be updated to require the corrected fdlibm.

Comments
CONVERTED DATA BugTraq+ Release Management Values COMMIT TO FIX: tiger-beta2 FIXED IN: tiger-beta2 INTEGRATED IN: tiger-b51 tiger-beta2
2004-06-14

SUGGESTED FIX src/share/classes/java/lang>sccs sccsdiff -r1.24 -r1.25 StrictMath.java ------- StrictMath.java ------- 28,40c28,39 < * <p>The Java math library is defined with respect to the version of < * <code>fdlibm</code> dated January 4, 1995. Where < * <code>fdlibm</code> provides more than one definition for a < * function (such as <code>acos</code>), use the "IEEE 754 core < * function" version (residing in a file whose name begins with the < * letter <code>e</code>). The methods which require < * <code>fdlibm</code> semantics are <code>sin</code>, < * <code>cos</code>, <code>tan</code>, <code>asin</code>, < * <code>acos</code>, <code>atan</code>, <code>exp</code>, < * <code>log</code>, <code>log10</code>, <code>cbrt</code>, < * <code>atan2</code>, <code>pow</code>, <code>sinh</code>, < * <code>cosh</code>, <code>tanh</code>, <code>hypot</code>, < * <code>expm1</code>, and <code>log1p</code>. --- > * <p>The Java math library is defined with respect to > * <code>fdlibm</code> version 5.3. Where <code>fdlibm</code> provides > * more than one definition for a function (such as > * <code>acos</code>), use the "IEEE 754 core function" version > * (residing in a file whose name begins with the letter > * <code>e</code>). The methods which require <code>fdlibm</code> > * semantics are <code>sin</code>, <code>cos</code>, <code>tan</code>, > * <code>asin</code>, <code>acos</code>, <code>atan</code>, > * <code>exp</code>, <code>log</code>, <code>log10</code>, > * <code>cbrt</code>, <code>atan2</code>, <code>pow</code>, > * <code>sinh</code>, <code>cosh</code>, <code>tanh</code>, > * <code>hypot</code>, <code>expm1</code>, and <code>log1p</code>. src/share/native/java/lang/fdlibm/src>sccs sccsdiff -r1.9 -r1.10 e_pow.c ------- e_pow.c ------- 103c103 < int i,j,k,yisint,n; --- > int i0,i1,i,j,k,yisint,n; 106a107 > i0 = ((*(int*)&one)>>29)^1; i1=1-i0; 174a176,177 > n = (hx>>31)+1; > 176c179 < if((((hx>>31)+1)|yisint)==0) return (x-x)/(x-x); --- > if((n|yisint)==0) return (x-x)/(x-x); 177a181,183 > s = one; /* s (sign of result -ve**odd) = -1 else = 1 */ > if((n|(yisint-1))==0) s = -one;/* (-ve)**(odd int) */ > 185,186c191,192 < if(ix<0x3fefffff) return (hy<0)? huge*huge:tiny*tiny; < if(ix>0x3ff00000) return (hy>0)? huge*huge:tiny*tiny; --- > if(ix<0x3fefffff) return (hy<0)? s*huge*huge:s*tiny*tiny; > if(ix>0x3ff00000) return (hy>0)? s*huge*huge:s*tiny*tiny; 189c195 < t = x-1; /* t has 20 trailing zeros */ --- > t = ax-one; /* t has 20 trailing zeros */ 197c203 < double s2,s_h,s_l,t_h,t_l; --- > double ss,s2,s_h,s_l,t_h,t_l; 211c217 < /* compute s = s_h+s_l = (x-1)/(x+1) or (x-1.5)/(x+1.5) */ --- > /* compute ss = s_h+s_l = (x-1)/(x+1) or (x-1.5)/(x+1.5) */ 214,215c220,221 < s = u*v; < s_h = s; --- > ss = u*v; > s_h = ss; 223c229 < s2 = s*s; --- > s2 = ss*ss; 225c231 < r += s_l*(s_h+s); --- > r += s_l*(s_h+ss); 230c236 < /* u+v = s*(1+...) */ --- > /* u+v = ss*(1+...) */ 232,233c238,239 < v = s_l*t_h+t_l*s; < /* 2/(3log2)*(s+...) */ --- > v = s_l*t_h+t_l*ss; > /* 2/(3log2)*(ss+...) */ 239c245 < /* log2(ax) = (s+..)*2/(3*log2) = n + dp_h + z_h + z_l */ --- > /* log2(ax) = (ss+..)*2/(3*log2) = n + dp_h + z_h + z_l */ 246,248d251 < s = one; /* s (sign of result -ve**odd) = -1 else = 1 */ < if((((hx>>31)+1)|(yisint-1))==0) s = -one;/* (-ve)**(odd int) */ < src/share/native/java/lang/fdlibm/src>sccs sccsdiff -r1.8 -r1.9 k_tan.c ------- k_tan.c ------- 80,85c80,100 < if(ix<0x3e300000) /* x < 2**-28 */ < {if((int)x==0) { /* generate inexact */ < if(((ix|__LO(x))|(iy+1))==0) return one/fabs(x); < else return (iy==1)? x: -one/x; < } < } --- > if(ix<0x3e300000) { /* x < 2**-28 */ > if((int)x==0) { /* generate inexact */ > if (((ix | __LO(x)) | (iy + 1)) == 0) > return one / fabs(x); > else { > if (iy == 1) > return x; > else { /* compute -1 / (x+y) carefully */ > double a, t; > > z = w = x + y; > __LO(z) = 0; > v = y - (z - x); > t = a = -one / w; > __LO(t) = 0; > s = one + t * z; > return t + a * (s + t * v); > } > } > } > } ###@###.### 2004-04-28
2004-04-28

EVALUATION Should be fixed in Tiger. ###@###.### 2004-04-18
2004-04-18