Refactor i386 libm code forcing underflow exceptions.

This patch refactors code in sysdeps/i386/fpu that forces underflow
exceptions to use common macros for that purpose as far as possible.
(Although some of the macros end up used in only one place, I think
it's cleanest to define all these macros together so that all the code
forcing underflow uses such macros.  Some more uses of such macros
will also be introduced when fixing remaining bugs about missing
underflow exceptions, and it would be possible to do further
refactoring of the macros in i386-math-asm.h to share more code by
using other macros internally.  Places that test for underflow by
examining the representation of the argument with integer operations,
rather that using floating-point comparisons on the argument or
result, are unchanged by this patch.)

Most of this code uses a macro MO to abstract away the differences
between PIC and non-PIC memory references to constants.  log1p
functions, however, hardcoded PIC conditionals for this.  Because the
common macros rely on the use of MO, I changed the log1p functions to
use the normal style here, and, for consistency, also made that change
to log1pl which is otherwise unaffected by this patch.

Tested for x86.

	* sysdeps/i386/fpu/i386-math-asm.h (DEFINE_LDBL_MIN): New macro.
	(FLT_CHECK_FORCE_UFLOW): Likewise.
	(DBL_CHECK_FORCE_UFLOW): Likewise.
	(FLT_CHECK_FORCE_UFLOW_NARROW): Likewise.
	(DBL_CHECK_FORCE_UFLOW_NARROW): Likewise.
	(LDBL_CHECK_FORCE_UFLOW_NONNEG_NAN): Likewise.
	(FLT_CHECK_FORCE_UFLOW_NONNAN): Likewise.
	(DBL_CHECK_FORCE_UFLOW_NONNAN): Likewise.
	(FLT_CHECK_FORCE_UFLOW_NONNEG): Likewise.
	(DBL_CHECK_FORCE_UFLOW_NONNEG): Likewise.
	(LDBL_CHECK_FORCE_UFLOW_NONNEG): Likewise.
	* sysdeps/i386/fpu/e_asin.S: Include <i386-math-asm.h>.
	(dbl_min): Replace with use of DEFINE_DBL_MIN.
	(__ieee754_asin): Use DBL_CHECK_FORCE_UFLOW.
	* sysdeps/i386/fpu/e_asinf.S: Include <i386-math-asm.h>.
	(flt_min): Replace with use of DEFINE_FLT_MIN.
	(__ieee754_asinf): Use FLT_CHECK_FORCE_UFLOW.
	* sysdeps/i386/fpu/e_atan2.S: Include <i386-math-asm.h>.
	(dbl_min): Replace with use of DEFINE_DBL_MIN.
	(__ieee754_atan2): Use DBL_CHECK_FORCE_UFLOW_NARROW.
	* sysdeps/i386/fpu/e_atan2f.S: Include <i386-math-asm.h>.
	(flt_min): Replace with use of DEFINE_FLT_MIN.
	(__ieee754_atan2f): Use FLT_CHECK_FORCE_UFLOW_NARROW.
	* sysdeps/i386/fpu/e_atanh.S: Include <i386-math-asm.h>.
	(dbl_min): Replace with use of DEFINE_DBL_MIN.
	(__ieee754_atanh): Use DBL_CHECK_FORCE_UFLOW_NONNEG.
	* sysdeps/i386/fpu/e_atanhf.S: Include <i386-math-asm.h>.
	(flt_min): Replace with use of DEFINE_FLT_MIN.
	(__ieee754_atanhf): Use FLT_CHECK_FORCE_UFLOW_NONNEG.
	* sysdeps/i386/fpu/e_exp2l.S: Include <i386-math-asm.h>.
	(ldbl_min): Replace with use of DEFINE_LDBL_MIN.
	(__ieee754_exp2l): Use LDBL_CHECK_FORCE_UFLOW_NONNEG_NAN.
	* sysdeps/i386/fpu/e_expl.S: Include <i386-math-asm.h>.
	[!USE_AS_EXPM1L] (cmin): Replace with use of DEFINE_LDBL_MIN.
	(IEEE754_EXPL): Use LDBL_CHECK_FORCE_UFLOW_NONNEG.
	* sysdeps/i386/fpu/s_atan.S: Include <i386-math-asm.h>.
	(dbl_min): Replace with use of DEFINE_DBL_MIN.
	(__atan): Use DBL_CHECK_FORCE_UFLOW.
	* sysdeps/i386/fpu/s_atanf.S: Include <i386-math-asm.h>.
	(flt_min): Replace with use of DEFINE_FLT_MIN.
	(__atanf): Use FLT_CHECK_FORCE_UFLOW.
	* sysdeps/i386/fpu/s_expm1.S: Include <i386-math-asm.h>.
	(dbl_min): Replace with use of DEFINE_DBL_MIN.
	(__expm1): Use DBL_CHECK_FORCE_UFLOW.  Move underflow check after
	main computation.
	* sysdeps/i386/fpu/s_expm1f.S: Include <i386-math-asm.h>.
	(flt_min): Replace with use of DEFINE_FLT_MIN.
	(__expm1f): Use FLT_CHECK_FORCE_UFLOW.  Move underflow check after
	main computation.
	* sysdeps/i386/fpu/s_log1p.S: Include <i386-math-asm.h>.
	(dbl_min): Replace with use of DEFINE_DBL_MIN.
	(MO): New macro.
	(__log1p): Use MO.  Use DBL_CHECK_FORCE_UFLOW_NONNAN.
	* sysdeps/i386/fpu/s_log1pf.S: Include <i386-math-asm.h>.
	(flt_min): Replace with use of DEFINE_FLT_MIN.
	(MO): New macro.
	(__log1pf): Use MO.  Use FLT_CHECK_FORCE_UFLOW_NONNAN.
	* sysdeps/i386/fpu/s_log1pl.S (MO): New macro.
	(__log1pl): Use MO.

ChangeLog

@@ -1,5 +1,65 @@
 2015-09-24  Joseph Myers  <joseph@codesourcery.com>
 
+	* sysdeps/i386/fpu/i386-math-asm.h (DEFINE_LDBL_MIN): New macro.
+	(FLT_CHECK_FORCE_UFLOW): Likewise.
+	(DBL_CHECK_FORCE_UFLOW): Likewise.
+	(FLT_CHECK_FORCE_UFLOW_NARROW): Likewise.
+	(DBL_CHECK_FORCE_UFLOW_NARROW): Likewise.
+	(LDBL_CHECK_FORCE_UFLOW_NONNEG_NAN): Likewise.
+	(FLT_CHECK_FORCE_UFLOW_NONNAN): Likewise.
+	(DBL_CHECK_FORCE_UFLOW_NONNAN): Likewise.
+	(FLT_CHECK_FORCE_UFLOW_NONNEG): Likewise.
+	(DBL_CHECK_FORCE_UFLOW_NONNEG): Likewise.
+	(LDBL_CHECK_FORCE_UFLOW_NONNEG): Likewise.
+	* sysdeps/i386/fpu/e_asin.S: Include <i386-math-asm.h>.
+	(dbl_min): Replace with use of DEFINE_DBL_MIN.
+	(__ieee754_asin): Use DBL_CHECK_FORCE_UFLOW.
+	* sysdeps/i386/fpu/e_asinf.S: Include <i386-math-asm.h>.
+	(flt_min): Replace with use of DEFINE_FLT_MIN.
+	(__ieee754_asinf): Use FLT_CHECK_FORCE_UFLOW.
+	* sysdeps/i386/fpu/e_atan2.S: Include <i386-math-asm.h>.
+	(dbl_min): Replace with use of DEFINE_DBL_MIN.
+	(__ieee754_atan2): Use DBL_CHECK_FORCE_UFLOW_NARROW.
+	* sysdeps/i386/fpu/e_atan2f.S: Include <i386-math-asm.h>.
+	(flt_min): Replace with use of DEFINE_FLT_MIN.
+	(__ieee754_atan2f): Use FLT_CHECK_FORCE_UFLOW_NARROW.
+	* sysdeps/i386/fpu/e_atanh.S: Include <i386-math-asm.h>.
+	(dbl_min): Replace with use of DEFINE_DBL_MIN.
+	(__ieee754_atanh): Use DBL_CHECK_FORCE_UFLOW_NONNEG.
+	* sysdeps/i386/fpu/e_atanhf.S: Include <i386-math-asm.h>.
+	(flt_min): Replace with use of DEFINE_FLT_MIN.
+	(__ieee754_atanhf): Use FLT_CHECK_FORCE_UFLOW_NONNEG.
+	* sysdeps/i386/fpu/e_exp2l.S: Include <i386-math-asm.h>.
+	(ldbl_min): Replace with use of DEFINE_LDBL_MIN.
+	(__ieee754_exp2l): Use LDBL_CHECK_FORCE_UFLOW_NONNEG_NAN.
+	* sysdeps/i386/fpu/e_expl.S: Include <i386-math-asm.h>.
+	[!USE_AS_EXPM1L] (cmin): Replace with use of DEFINE_LDBL_MIN.
+	(IEEE754_EXPL): Use LDBL_CHECK_FORCE_UFLOW_NONNEG.
+	* sysdeps/i386/fpu/s_atan.S: Include <i386-math-asm.h>.
+	(dbl_min): Replace with use of DEFINE_DBL_MIN.
+	(__atan): Use DBL_CHECK_FORCE_UFLOW.
+	* sysdeps/i386/fpu/s_atanf.S: Include <i386-math-asm.h>.
+	(flt_min): Replace with use of DEFINE_FLT_MIN.
+	(__atanf): Use FLT_CHECK_FORCE_UFLOW.
+	* sysdeps/i386/fpu/s_expm1.S: Include <i386-math-asm.h>.
+	(dbl_min): Replace with use of DEFINE_DBL_MIN.
+	(__expm1): Use DBL_CHECK_FORCE_UFLOW.  Move underflow check after
+	main computation.
+	* sysdeps/i386/fpu/s_expm1f.S: Include <i386-math-asm.h>.
+	(flt_min): Replace with use of DEFINE_FLT_MIN.
+	(__expm1f): Use FLT_CHECK_FORCE_UFLOW.  Move underflow check after
+	main computation.
+	* sysdeps/i386/fpu/s_log1p.S: Include <i386-math-asm.h>.
+	(dbl_min): Replace with use of DEFINE_DBL_MIN.
+	(MO): New macro.
+	(__log1p): Use MO.  Use DBL_CHECK_FORCE_UFLOW_NONNAN.
+	* sysdeps/i386/fpu/s_log1pf.S: Include <i386-math-asm.h>.
+	(flt_min): Replace with use of DEFINE_FLT_MIN.
+	(MO): New macro.
+	(__log1pf): Use MO.  Use FLT_CHECK_FORCE_UFLOW_NONNAN.
+	* sysdeps/i386/fpu/s_log1pl.S (MO): New macro.
+	(__log1pl): Use MO.
+
 	[BZ #19003]
 	* sysdeps/x86_64/fpu/multiarch/Makefile (CFLAGS-e_pow-fma4.c): Add
 	$(config-cflags-nofma).

sysdeps/i386/fpu/e_asin.S

@@ -4,15 +4,11 @@
  */
 
 #include <machine/asm.h>
+#include <i386-math-asm.h>
 
 RCSID("$NetBSD: e_asin.S,v 1.4 1995/05/08 23:45:40 jtc Exp $")
 
-	.section .rodata.cst8,"aM",@progbits,8
+DEFINE_DBL_MIN
-
-	.p2align 3
-	.type dbl_min,@object
-dbl_min:	.byte 0, 0, 0, 0, 0, 0, 0x10, 0
-	ASM_SIZE_DIRECTIVE(dbl_min)
 
 #ifdef PIC
 # define MO(op) op##@GOTOFF(%ecx)
@@ -36,20 +32,7 @@ ENTRY(__ieee754_asin)
 	fmulp				/* 1 - x^2 */
 	fsqrt				/* sqrt (1 - x^2) */
 	fpatan
-	fldl	MO(dbl_min)
+	DBL_CHECK_FORCE_UFLOW
-	fld	%st(1)
+	ret
-	fabs
-	fucompp
-	fnstsw
-	sahf
-	jnc	1f
-	subl	$8, %esp
-	cfi_adjust_cfa_offset (8)
-	fld	%st(0)
-	fmul	%st(0)
-	fstpl	(%esp)
-	addl	$8, %esp
-	cfi_adjust_cfa_offset (-8)
-1:	ret
 END (__ieee754_asin)
 strong_alias (__ieee754_asin, __asin_finite)

sysdeps/i386/fpu/e_asinf.S

@@ -5,15 +5,13 @@
  */
 
 #include <machine/asm.h>
+#include <i386-math-asm.h>
 
 RCSID("$NetBSD: $")
 
 	.section .rodata.cst4,"aM",@progbits,4
 
-	.p2align 2
+DEFINE_FLT_MIN
-	.type flt_min,@object
-flt_min:	.byte 0, 0, 0x80, 0
-	ASM_SIZE_DIRECTIVE(flt_min)
 
 #ifdef PIC
 # define MO(op) op##@GOTOFF(%ecx)
@@ -35,20 +33,7 @@ ENTRY(__ieee754_asinf)
 	fsubp				/* 1 - x^2 */
 	fsqrt				/* sqrt (1 - x^2) */
 	fpatan
-	flds	MO(flt_min)
+	FLT_CHECK_FORCE_UFLOW
-	fld	%st(1)
+	ret
-	fabs
-	fucompp
-	fnstsw
-	sahf
-	jnc	1f
-	subl	$4, %esp
-	cfi_adjust_cfa_offset (4)
-	fld	%st(0)
-	fmul	%st(0)
-	fstps	(%esp)
-	addl	$4, %esp
-	cfi_adjust_cfa_offset (-4)
-1:	ret
 END (__ieee754_asinf)
 strong_alias (__ieee754_asinf, __asinf_finite)

sysdeps/i386/fpu/e_atan2.S

@@ -4,15 +4,11 @@
  */
 
 #include <machine/asm.h>
+#include <i386-math-asm.h>
 
 RCSID("$NetBSD: e_atan2.S,v 1.4 1995/05/08 23:46:28 jtc Exp $")
 
-	.section .rodata.cst8,"aM",@progbits,8
+DEFINE_DBL_MIN
-
-	.p2align 3
-	.type dbl_min,@object
-dbl_min:	.byte 0, 0, 0, 0, 0, 0, 0x10, 0
-	ASM_SIZE_DIRECTIVE(dbl_min)
 
 #ifdef PIC
 # define MO(op) op##@GOTOFF(%ecx)
@@ -28,22 +24,7 @@ ENTRY(__ieee754_atan2)
 	fldl	 4(%esp)
 	fldl	12(%esp)
 	fpatan
-	fldl	MO(dbl_min)
+	DBL_CHECK_FORCE_UFLOW_NARROW
-	fld	%st(1)
+	ret
-	fabs
-	fucompp
-	fnstsw
-	sahf
-	jnc 1f
-	subl	$8, %esp
-	cfi_adjust_cfa_offset (8)
-	fld	%st(0)
-	fmul	%st(0)
-	fstpl	(%esp)
-	fstpl	(%esp)
-	fldl	(%esp)
-	addl	$8, %esp
-	cfi_adjust_cfa_offset (-8)
-1:	ret
 END (__ieee754_atan2)
 strong_alias (__ieee754_atan2, __atan2_finite)

sysdeps/i386/fpu/e_atan2f.S

@@ -4,15 +4,11 @@
  */
 
 #include <machine/asm.h>
+#include <i386-math-asm.h>
 
 RCSID("$NetBSD: e_atan2f.S,v 1.1 1995/05/08 23:35:10 jtc Exp $")
 
-	.section .rodata.cst4,"aM",@progbits,4
+DEFINE_FLT_MIN
-
-	.p2align 2
-	.type flt_min,@object
-flt_min:	.byte 0, 0, 0x80, 0
-	ASM_SIZE_DIRECTIVE(flt_min)
 
 #ifdef PIC
 # define MO(op) op##@GOTOFF(%ecx)
@@ -28,22 +24,7 @@ ENTRY(__ieee754_atan2f)
 	flds	4(%esp)
 	flds	8(%esp)
 	fpatan
-	flds	MO(flt_min)
+	FLT_CHECK_FORCE_UFLOW_NARROW
-	fld	%st(1)
+	ret
-	fabs
-	fucompp
-	fnstsw
-	sahf
-	jnc 1f
-	subl	$4, %esp
-	cfi_adjust_cfa_offset (4)
-	fld	%st(0)
-	fmul	%st(0)
-	fstps	(%esp)
-	fstps	(%esp)
-	flds	(%esp)
-	addl	$4, %esp
-	cfi_adjust_cfa_offset (-4)
-1:	ret
 END (__ieee754_atan2f)
 strong_alias (__ieee754_atan2f, __atan2f_finite)

sysdeps/i386/fpu/e_atanh.S

@@ -18,6 +18,7 @@
    <http://www.gnu.org/licenses/>.  */
 
 #include <machine/asm.h>
+#include <i386-math-asm.h>
 
 	.section .rodata
 
@@ -35,12 +36,7 @@ limit:	.double 0.29
 ln2_2:	.tfloat 0.3465735902799726547086160
 	ASM_SIZE_DIRECTIVE(ln2_2)
 
-	.section .rodata.cst8,"aM",@progbits,8
+DEFINE_DBL_MIN
-
-	.p2align 3
-	.type dbl_min,@object
-dbl_min:	.byte 0, 0, 0, 0, 0, 0, 0x10, 0
-	ASM_SIZE_DIRECTIVE(dbl_min)
 
 #ifdef PIC
 #define MO(op) op##@GOTOFF(%edx)
@@ -88,18 +84,8 @@ ENTRY(__ieee754_atanh)
 	sahf
 	jae	4f
 	fyl2xp1			// 0.5*ln2*ld(1+2*|x|+(2*|x|^2)/(1-|x|))
-	fcoml	MO(dbl_min)
+	DBL_CHECK_FORCE_UFLOW_NONNEG
-	fnstsw
+	jecxz	3f
-	sahf
-	jae	8f
-	subl	$8, %esp
-	cfi_adjust_cfa_offset (8)
-	fld	%st(0)
-	fmul	%st(0)
-	fstpl	(%esp)
-	addl	$8, %esp
-	cfi_adjust_cfa_offset (-8)
-8:	jecxz	3f
 	fchs			// 0.5*ln2*ld(1+2*x+(2*x^2)/(1-x))
 3:	ret
 

sysdeps/i386/fpu/e_atanhf.S

@@ -18,6 +18,7 @@
    <http://www.gnu.org/licenses/>.  */
 
 #include <machine/asm.h>
+#include <i386-math-asm.h>
 
 	.section .rodata
 
@@ -36,12 +37,7 @@ limit:	.double 0.29
 ln2_2:	.tfloat 0.3465735902799726547086160
 	ASM_SIZE_DIRECTIVE(ln2_2)
 
-	.section .rodata.cst4,"aM",@progbits,4
+DEFINE_FLT_MIN
-
-	.p2align 2
-	.type flt_min,@object
-flt_min:	.byte 0, 0, 0x80, 0
-	ASM_SIZE_DIRECTIVE(flt_min)
 
 #ifdef PIC
 #define MO(op) op##@GOTOFF(%edx)
@@ -84,18 +80,8 @@ ENTRY(__ieee754_atanhf)
 	sahf
 	jae	4f
 	fyl2xp1			// 0.5*ln2*ld(1+2*|x|+(2*|x|^2)/(1-|x|))
-	fcoms	MO(flt_min)
+	FLT_CHECK_FORCE_UFLOW_NONNEG
-	fnstsw
+	jecxz	3f
-	sahf
-	jae	6f
-	subl	$4, %esp
-	cfi_adjust_cfa_offset (4)
-	fld	%st(0)
-	fmul	%st(0)
-	fstps	(%esp)
-	addl	$4, %esp
-	cfi_adjust_cfa_offset (-4)
-6:	jecxz	3f
 	fchs			// 0.5*ln2*ld(1+2*x+(2*x^2)/(1-x))
 3:	ret
 

sysdeps/i386/fpu/e_exp2l.S

@@ -5,13 +5,9 @@
  */
 
 #include <machine/asm.h>
+#include <i386-math-asm.h>
 
-	.section .rodata.cst16,"aM",@progbits,16
+DEFINE_LDBL_MIN
-	.p2align 4
-	.type ldbl_min,@object
-ldbl_min:	.byte 0, 0, 0, 0, 0, 0, 0, 0x80, 0x1, 0
-	.byte 0, 0, 0, 0, 0, 0
-	ASM_SIZE_DIRECTIVE(ldbl_min)
 
 #ifdef PIC
 # define MO(op) op##@GOTOFF(%ecx)
@@ -52,17 +48,8 @@ ENTRY(__ieee754_exp2l)
 	faddp				/* 2^(fract(x)) */
 	fscale				/* e^x */
 	fstp	%st(1)
-	/* Ensure underflow for tiny result.  */
+	LDBL_CHECK_FORCE_UFLOW_NONNEG_NAN
-	fldt	MO(ldbl_min)
+	ret
-	fld	%st(1)
-	fucompp
-	fnstsw
-	sahf
-	jnc	4f
-	fld	%st
-	fmul	%st
-	fstp	%st
-4:	ret
 
 1:	testl	$0x200, %eax		/* Test sign.  */
 	jz	2f			/* If positive, jump.  */

sysdeps/i386/fpu/e_expl.S

@@ -23,6 +23,7 @@
  */
 
 #include <machine/asm.h>
+#include <i386-math-asm.h>
 
 #ifdef USE_AS_EXP10L
 # define IEEE754_EXPL __ieee754_exp10l
@@ -65,10 +66,7 @@ c1:	.byte 0x20, 0xfa, 0xee, 0xc2, 0x5f, 0x70, 0xa5, 0xec, 0xed, 0x3f
 csat:	.byte 0, 0, 0, 0, 0, 0, 0, 0x80, 0x0e, 0x40
 	.byte 0, 0, 0, 0, 0, 0
 	ASM_SIZE_DIRECTIVE(csat)
-	.type cmin,@object
+DEFINE_LDBL_MIN
-cmin:	.byte 0, 0, 0, 0, 0, 0, 0, 0x80, 0x1, 0
-	.byte 0, 0, 0, 0, 0, 0
-	ASM_SIZE_DIRECTIVE(cmin)
 #endif
 
 #ifdef PIC
@@ -199,18 +197,9 @@ ENTRY(IEEE754_EXPL)
 	fstp	%st(1)		/* 2  */
 	fscale			/* 2 scale factor is st(1); base^x */
 	fstp	%st(1)		/* 1  */
-	/* Ensure underflow for tiny result.  */
+	LDBL_CHECK_FORCE_UFLOW_NONNEG
-	fldt	MO(cmin)	/* 2 cmin  */
-	fld	%st(1)		/* 3  */
-	fcompp			/* 1  */
-	fnstsw
-	sahf
-	jnc	6f
-	fld	%st
-	fmul	%st
-	fstp	%st
 #endif
-6:	fstp	%st(1)		/* 0  */
+	fstp	%st(1)		/* 0  */
 	jmp	2f
 1:
 #ifdef USE_AS_EXPM1L

sysdeps/i386/fpu/i386-math-asm.h

@@ -52,6 +52,14 @@ flt_min:					\
 dbl_min:					\
 	.byte 0, 0, 0, 0, 0, 0, 0x10, 0;	\
 	.size dbl_min, .-dbl_min;
+#define DEFINE_LDBL_MIN					\
+	.section .rodata.cst16,"aM",@progbits,16;	\
+	.p2align 4;					\
+	.type ldbl_min,@object;				\
+ldbl_min:						\
+	.byte 0, 0, 0, 0, 0, 0, 0, 0x80, 0x1, 0;	\
+	.byte 0, 0, 0, 0, 0, 0;				\
+	.size ldbl_min, .-ldbl_min;
 
 /* Remove excess range and precision by storing a value on the stack
    and loading it back.  The value is given to be nonnegative or NaN;
@@ -123,4 +131,164 @@ dbl_min:					\
 	addl	$8, %esp;			\
 	cfi_adjust_cfa_offset (-8);
 
+/* Force an underflow exception if the given value is subnormal.  The
+   relevant constant for the minimum of the type must have been
+   defined, the MO macro must have been defined for access to memory
+   operands, and, if PIC, the PIC register must have been loaded.  */
+#define FLT_CHECK_FORCE_UFLOW			\
+	flds	MO(flt_min);			\
+	fld	%st(1);				\
+	fabs;					\
+	fucompp;				\
+	fnstsw;					\
+	sahf;					\
+	jnc 6424f;				\
+	subl	$4, %esp;			\
+	cfi_adjust_cfa_offset (4);		\
+	fld	%st(0);				\
+	fmul	%st(0);				\
+	fstps	(%esp);				\
+	addl	$4, %esp;			\
+	cfi_adjust_cfa_offset (-4);		\
+6424:
+#define DBL_CHECK_FORCE_UFLOW			\
+	fldl	MO(dbl_min);			\
+	fld	%st(1);				\
+	fabs;					\
+	fucompp;				\
+	fnstsw;					\
+	sahf;					\
+	jnc 6453f;				\
+	subl	$8, %esp;			\
+	cfi_adjust_cfa_offset (8);		\
+	fld	%st(0);				\
+	fmul	%st(0);				\
+	fstpl	(%esp);				\
+	addl	$8, %esp;			\
+	cfi_adjust_cfa_offset (-8);		\
+6453:
+
+/* Likewise, but also remove excess range and precision if the value
+   is subnormal.  */
+#define FLT_CHECK_FORCE_UFLOW_NARROW		\
+	flds	MO(flt_min);			\
+	fld	%st(1);				\
+	fabs;					\
+	fucompp;				\
+	fnstsw;					\
+	sahf;					\
+	jnc 6424f;				\
+	subl	$4, %esp;			\
+	cfi_adjust_cfa_offset (4);		\
+	fld	%st(0);				\
+	fmul	%st(0);				\
+	fstps	(%esp);				\
+	fstps	(%esp);				\
+	flds	(%esp);				\
+	addl	$4, %esp;			\
+	cfi_adjust_cfa_offset (-4);		\
+6424:
+#define DBL_CHECK_FORCE_UFLOW_NARROW		\
+	fldl	MO(dbl_min);			\
+	fld	%st(1);				\
+	fabs;					\
+	fucompp;				\
+	fnstsw;					\
+	sahf;					\
+	jnc 6453f;				\
+	subl	$8, %esp;			\
+	cfi_adjust_cfa_offset (8);		\
+	fld	%st(0);				\
+	fmul	%st(0);				\
+	fstpl	(%esp);				\
+	fstpl	(%esp);				\
+	fldl	(%esp);				\
+	addl	$8, %esp;			\
+	cfi_adjust_cfa_offset (-8);		\
+6453:
+
+/* Likewise, but the argument is nonnegative or NaN.  */
+#define LDBL_CHECK_FORCE_UFLOW_NONNEG_NAN	\
+	fldt	MO(ldbl_min);			\
+	fld	%st(1);				\
+	fucompp;				\
+	fnstsw;					\
+	sahf;					\
+	jnc 6464f;				\
+	fld	%st(0);				\
+	fmul	%st(0);				\
+	fstp	%st(0);				\
+6464:
+
+/* Likewise, but the argument is not a NaN.  */
+#define FLT_CHECK_FORCE_UFLOW_NONNAN		\
+	fld %st(0);				\
+	fabs;					\
+	fcomps	MO(flt_min);			\
+	fnstsw;					\
+	sahf;					\
+	jnc 6424f;				\
+	subl	$4, %esp;			\
+	cfi_adjust_cfa_offset (4);		\
+	fld	%st(0);				\
+	fmul	%st(0);				\
+	fstps	(%esp);				\
+	addl	$4, %esp;			\
+	cfi_adjust_cfa_offset (-4);		\
+6424:
+#define DBL_CHECK_FORCE_UFLOW_NONNAN		\
+	fld %st(0);				\
+	fabs;					\
+	fcompl	MO(dbl_min);			\
+	fnstsw;					\
+	sahf;					\
+	jnc 6453f;				\
+	subl	$8, %esp;			\
+	cfi_adjust_cfa_offset (8);		\
+	fld	%st(0);				\
+	fmul	%st(0);				\
+	fstpl	(%esp);				\
+	addl	$8, %esp;			\
+	cfi_adjust_cfa_offset (-8);		\
+6453:
+
+/* Likewise, but the argument is nonnegative and not a NaN.  */
+#define FLT_CHECK_FORCE_UFLOW_NONNEG		\
+	fcoms	MO(flt_min);			\
+	fnstsw;					\
+	sahf;					\
+	jnc 6424f;				\
+	subl	$4, %esp;			\
+	cfi_adjust_cfa_offset (4);		\
+	fld	%st(0);				\
+	fmul	%st(0);				\
+	fstps	(%esp);				\
+	addl	$4, %esp;			\
+	cfi_adjust_cfa_offset (-4);		\
+6424:
+#define DBL_CHECK_FORCE_UFLOW_NONNEG		\
+	fcoml	MO(dbl_min);			\
+	fnstsw;					\
+	sahf;					\
+	jnc 6453f;				\
+	subl	$8, %esp;			\
+	cfi_adjust_cfa_offset (8);		\
+	fld	%st(0);				\
+	fmul	%st(0);				\
+	fstpl	(%esp);				\
+	addl	$8, %esp;			\
+	cfi_adjust_cfa_offset (-8);		\
+6453:
+#define LDBL_CHECK_FORCE_UFLOW_NONNEG		\
+	fldt	MO(ldbl_min);			\
+	fld	%st(1);				\
+	fcompp;					\
+	fnstsw;					\
+	sahf;					\
+	jnc 6464f;				\
+	fld	%st(0);				\
+	fmul	%st(0);				\
+	fstp	%st(0);				\
+6464:
+
 #endif /* i386-math-asm.h.  */

sysdeps/i386/fpu/s_atan.S

@@ -4,15 +4,11 @@
  */
 
 #include <machine/asm.h>
+#include <i386-math-asm.h>
 
 RCSID("$NetBSD: s_atan.S,v 1.4 1995/05/08 23:50:41 jtc Exp $")
 
-	.section .rodata.cst8,"aM",@progbits,8
+DEFINE_DBL_MIN
-
-	.p2align 3
-	.type dbl_min,@object
-dbl_min:	.byte 0, 0, 0, 0, 0, 0, 0x10, 0
-	ASM_SIZE_DIRECTIVE(dbl_min)
 
 #ifdef PIC
 # define MO(op) op##@GOTOFF(%ecx)
@@ -28,20 +24,7 @@ ENTRY(__atan)
 	fldl	4(%esp)
 	fld1
 	fpatan
-	fldl	MO(dbl_min)
+	DBL_CHECK_FORCE_UFLOW
-	fld	%st(1)
+	ret
-	fabs
-	fucompp
-	fnstsw
-	sahf
-	jnc 1f
-	subl	$8, %esp
-	cfi_adjust_cfa_offset (8)
-	fld	%st(0)
-	fmul	%st(0)
-	fstpl	(%esp)
-	addl	$8, %esp
-	cfi_adjust_cfa_offset (-8)
-1:	ret
 END (__atan)
 weak_alias (__atan, atan)

sysdeps/i386/fpu/s_atanf.S

@@ -4,15 +4,11 @@
  */
 
 #include <machine/asm.h>
+#include <i386-math-asm.h>
 
 RCSID("$NetBSD: s_atanf.S,v 1.3 1995/05/08 23:51:33 jtc Exp $")
 
-	.section .rodata.cst4,"aM",@progbits,4
+DEFINE_FLT_MIN
-
-	.p2align 2
-	.type flt_min,@object
-flt_min:	.byte 0, 0, 0x80, 0
-	ASM_SIZE_DIRECTIVE(flt_min)
 
 #ifdef PIC
 # define MO(op) op##@GOTOFF(%ecx)
@@ -28,20 +24,7 @@ ENTRY(__atanf)
 	flds	4(%esp)
 	fld1
 	fpatan
-	flds	MO(flt_min)
+	FLT_CHECK_FORCE_UFLOW
-	fld	%st(1)
+	ret
-	fabs
-	fucompp
-	fnstsw
-	sahf
-	jnc 1f
-	subl	$4, %esp
-	cfi_adjust_cfa_offset (4)
-	fld	%st(0)
-	fmul	%st(0)
-	fstps	(%esp)
-	addl	$4, %esp
-	cfi_adjust_cfa_offset (-4)
-1:	ret
 END (__atanf)
 weak_alias (__atanf, atanf)

sysdeps/i386/fpu/s_expm1.S

@@ -23,6 +23,7 @@
 
 #include <sysdep.h>
 #include <machine/asm.h>
+#include <i386-math-asm.h>
 
 	.section .rodata
 
@@ -37,12 +38,7 @@ one:	.double 1.0
 l2e:	.tfloat 1.442695040888963407359924681002
 	ASM_SIZE_DIRECTIVE(l2e)
 
-	.section .rodata.cst8,"aM",@progbits,8
+DEFINE_DBL_MIN
-
-	.p2align 3
-	.type dbl_min,@object
-dbl_min:	.byte 0, 0, 0, 0, 0, 0, 0x10, 0
-	ASM_SIZE_DIRECTIVE(dbl_min)
 
 #ifdef PIC
 #define MO(op) op##@GOTOFF(%edx)
@@ -81,21 +77,6 @@ ENTRY(__expm1)
 #ifdef	PIC
 	LOAD_PIC_REG (dx)
 #endif
-	fld	%st
-	fabs
-	fcoml	MO(dbl_min)
-	fstp	%st
-	fnstsw
-	sahf
-	jae	5f
-	subl	$8, %esp
-	cfi_adjust_cfa_offset (8)
-	fld	%st(0)
-	fmul	%st(0)
-	fstpl	(%esp)
-	addl	$8, %esp
-	cfi_adjust_cfa_offset (-8)
-	ret
 
 5:	fldt	MO(l2e)		// log2(e) : x
 	fmulp			// log2(e)*x
@@ -122,6 +103,7 @@ ENTRY(__expm1)
 	fsubrl	MO(one)		// 1-2^int(log2(e)*x) : int(log2(e)*x) : 2^(log2(e)*x)-2^int(log2(e)*x)
 	fstp	%st(1)		// 1-2^int(log2(e)*x) : 2^(log2(e)*x)-2^int(log2(e)*x)
 	fsubrp	%st, %st(1)	// 2^(log2(e)*x)
+	DBL_CHECK_FORCE_UFLOW
 	ret
 
 2:	fstp	%st

sysdeps/i386/fpu/s_expm1f.S

@@ -23,6 +23,7 @@
 
 #include <sysdep.h>
 #include <machine/asm.h>
+#include <i386-math-asm.h>
 
 	.section .rodata
 
@@ -37,12 +38,7 @@ one:	.double 1.0
 l2e:	.tfloat 1.442695040888963407359924681002
 	ASM_SIZE_DIRECTIVE(l2e)
 
-	.section .rodata.cst4,"aM",@progbits,4
+DEFINE_FLT_MIN
-
-	.p2align 2
-	.type flt_min,@object
-flt_min:	.byte 0, 0, 0x80, 0
-	ASM_SIZE_DIRECTIVE(flt_min)
 
 #ifdef PIC
 #define MO(op) op##@GOTOFF(%edx)
@@ -81,21 +77,6 @@ ENTRY(__expm1f)
 #ifdef	PIC
 	LOAD_PIC_REG (dx)
 #endif
-	fld	%st
-	fabs
-	fcoms	MO(flt_min)
-	fstp	%st
-	fnstsw
-	sahf
-	jae	5f
-	subl	$4, %esp
-	cfi_adjust_cfa_offset (4)
-	fld	%st(0)
-	fmul	%st(0)
-	fstps	(%esp)
-	addl	$4, %esp
-	cfi_adjust_cfa_offset (-4)
-	ret
 
 5:	fldt	MO(l2e)		// log2(e) : x
 	fmulp			// log2(e)*x
@@ -122,6 +103,7 @@ ENTRY(__expm1f)
 	fsubrl	MO(one)		// 1-2^int(log2(e)*x) : int(log2(e)*x) : 2^(log2(e)*x)-2^int(log2(e)*x)
 	fstp	%st(1)		// 1-2^int(log2(e)*x) : 2^(log2(e)*x)-2^int(log2(e)*x)
 	fsubrp	%st, %st(1)	// 2^(log2(e)*x)
+	FLT_CHECK_FORCE_UFLOW
 	ret
 
 2:	fstp	%st

sysdeps/i386/fpu/s_log1p.S

@@ -4,6 +4,7 @@
  */
 
 #include <machine/asm.h>
+#include <i386-math-asm.h>
 
 RCSID("$NetBSD: s_log1p.S,v 1.7 1995/05/09 00:10:58 jtc Exp $")
 
@@ -17,12 +18,13 @@ RCSID("$NetBSD: s_log1p.S,v 1.7 1995/05/09 00:10:58 jtc Exp $")
 limit:	.double 0.29
 one:	.double 1.0
 
-	.section .rodata.cst8,"aM",@progbits,8
+DEFINE_DBL_MIN
 
-	.p2align 3
+#ifdef PIC
-	.type dbl_min,@object
+# define MO(op) op##@GOTOFF(%edx)
-dbl_min:	.byte 0, 0, 0, 0, 0, 0, 0x10, 0
+#else
-	ASM_SIZE_DIRECTIVE(dbl_min)
+# define MO(op) op
+#endif
 
 /*
  * Use the fyl2xp1 function when the argument is in the range -0.29 to 0.29,
@@ -44,43 +46,18 @@ ENTRY(__log1p)
 	sahf
 	jc	3f		// in case x is NaN or ±Inf
 4:	fabs
-#ifdef PIC
+	fcompl	MO(limit)
-	fcompl	limit@GOTOFF(%edx)
-#else
-	fcompl	limit
-#endif
 	fnstsw
 	sahf
 	jc	2f
 
-#ifdef PIC
+	faddl	MO(one)
-	faddl	one@GOTOFF(%edx)
-#else
-	faddl	one
-#endif
 	fyl2x
 	ret
 
 2:	fyl2xp1
-#ifdef PIC
+	DBL_CHECK_FORCE_UFLOW_NONNAN
-	fldl	dbl_min@GOTOFF(%edx)
+	ret
-#else
-	fldl	dbl_min
-#endif
-	fld	%st(1)
-	fabs
-	fucompp
-	fnstsw
-	sahf
-	jnc	1f
-	subl	$8, %esp
-	cfi_adjust_cfa_offset (8)
-	fld	%st(0)
-	fmul	%st(0)
-	fstpl	(%esp)
-	addl	$8, %esp
-	cfi_adjust_cfa_offset (-8)
-1:	ret
 
 3:	jp	4b		// in case x is ±Inf
 	fstp	%st(1)

sysdeps/i386/fpu/s_log1pf.S

@@ -4,6 +4,7 @@
  */
 
 #include <machine/asm.h>
+#include <i386-math-asm.h>
 
 RCSID("$NetBSD: s_log1pf.S,v 1.4 1995/05/09 00:13:05 jtc Exp $")
 
@@ -17,12 +18,13 @@ RCSID("$NetBSD: s_log1pf.S,v 1.4 1995/05/09 00:13:05 jtc Exp $")
 limit:	.float 0.29
 one:	.float 1.0
 
-	.section .rodata.cst4,"aM",@progbits,4
+DEFINE_FLT_MIN
 
-	.p2align 2
+#ifdef PIC
-	.type flt_min,@object
+# define MO(op) op##@GOTOFF(%edx)
-flt_min:	.byte 0, 0, 0x80, 0
+#else
-	ASM_SIZE_DIRECTIVE(flt_min)
+# define MO(op) op
+#endif
 
 /*
  * Use the fyl2xp1 function when the argument is in the range -0.29 to 0.29,
@@ -44,43 +46,18 @@ ENTRY(__log1pf)
 	sahf
 	jc	3f		// in case x is NaN or ±Inf
 4:	fabs
-#ifdef PIC
+	fcomps	MO(limit)
-	fcomps	limit@GOTOFF(%edx)
-#else
-	fcomps	limit
-#endif
 	fnstsw
 	sahf
 	jc	2f
 
-#ifdef PIC
+	fadds	MO(one)
-	fadds	one@GOTOFF(%edx)
-#else
-	fadds	one
-#endif
 	fyl2x
 	ret
 
 2:	fyl2xp1
-#ifdef PIC
+	FLT_CHECK_FORCE_UFLOW_NONNAN
-	flds	flt_min@GOTOFF(%edx)
+	ret
-#else
-	flds	flt_min
-#endif
-	fld	%st(1)
-	fabs
-	fucompp
-	fnstsw
-	sahf
-	jnc	1f
-	subl	$4, %esp
-	cfi_adjust_cfa_offset (4)
-	fld	%st(0)
-	fmul	%st(0)
-	fstps	(%esp)
-	addl	$4, %esp
-	cfi_adjust_cfa_offset (-4)
-1:	ret
 
 3:	jp	4b		// in case x is ±Inf
 	fstp	%st(1)

sysdeps/i386/fpu/s_log1pl.S

@@ -22,6 +22,12 @@ limit:	.tfloat 0.29
 	   but it helps to optimize the code.  */
 one:	.double 1.0
 
+#ifdef PIC
+# define MO(op) op##@GOTOFF(%edx)
+#else
+# define MO(op) op
+#endif
+
 /*
  * Use the fyl2xp1 function when the argument is in the range -0.29 to 0.29,
  * otherwise fyl2x with the needed extra computation.
@@ -43,11 +49,7 @@ ENTRY(__log1pl)
 	jc	3f		// in case x is NaN or ±Inf
 4:
 	fabs
-#ifdef PIC
+	fldt	MO(limit)
-	fldt	limit@GOTOFF(%edx)
-#else
-	fldt	limit
-#endif
 	fcompp
 	fnstsw
 	sahf
@@ -58,11 +60,7 @@ ENTRY(__log1pl)
 	cmpl	$0xc040, %eax
 	jae	5f
 
-#ifdef PIC
+	faddl	MO(one)
-	faddl	one@GOTOFF(%edx)
-#else
-	faddl	one
-#endif
 5:	fyl2x
 	ret