diff options
| author | Francois-Xavier Coudert <fxcoudert@gcc.gnu.org> | 2008-02-27 14:32:02 +0000 |
|---|---|---|
| committer | François-Xavier Coudert <fxcoudert@gcc.gnu.org> | 2008-02-27 14:32:02 +0000 |
| commit | b5a4419cde676a67cd150feeef3babea43e5af8c (patch) | |
| tree | 64de6d9997d0e3e22c4eeefedf20b872963e2089 /libjava/classpath/java | |
| parent | 8bf6e2702881e7d34ee9b127f716d30fc6ee0adf (diff) | |
| download | gcc-b5a4419cde676a67cd150feeef3babea43e5af8c.zip gcc-b5a4419cde676a67cd150feeef3babea43e5af8c.tar.gz gcc-b5a4419cde676a67cd150feeef3babea43e5af8c.tar.bz2 | |
re PR fortran/33387 (Fortran front-end should translate intrinsics by calling C99 function instead of libgfortran functions)
PR fortran/33387
* trans.h: Remove prototypes for gfor_fndecl_math_exponent4,
gfor_fndecl_math_exponent8, gfor_fndecl_math_exponent10 and
gfor_fndecl_math_exponent16.
* f95-lang.c (build_builtin_fntypes): Add new function types.
(gfc_init_builtin_functions): Add new builtins for nextafter,
frexp, ldexp, fabs, scalbn and inf.
* iresolve.c (gfc_resolve_rrspacing): Don't add hidden arguments.
(gfc_resolve_scale): Don't convert type of second argument.
(gfc_resolve_set_exponent): Likewise.
(gfc_resolve_size): Don't add hidden arguments.
* trans-decl.c: Remove gfor_fndecl_math_exponent4,
gfor_fndecl_math_exponent8, gfor_fndecl_math_exponent10 and
gfor_fndecl_math_exponent16.
* trans-intrinsic.c (gfc_intrinsic_map): Remove intrinsics
for scalbn, fraction, nearest, rrspacing, set_exponent and
spacing.
(gfc_conv_intrinsic_exponent): Directly call frexp.
(gfc_conv_intrinsic_fraction, gfc_conv_intrinsic_nearest,
gfc_conv_intrinsic_spacing, gfc_conv_intrinsic_rrspacing,
gfc_conv_intrinsic_scale, gfc_conv_intrinsic_set_exponent): New
functions.
(gfc_conv_intrinsic_function): Use the new functions above.
From-SVN: r132713
Diffstat (limited to 'libjava/classpath/java')
0 files changed, 0 insertions, 0 deletions