diff options
Diffstat (limited to 'src')
-rw-r--r-- | src/math/i386/asin.s | 23 | ||||
-rw-r--r-- | src/math/i386/atan.s | 10 | ||||
-rw-r--r-- | src/math/i386/atanf.s | 12 | ||||
-rw-r--r-- | src/math/i386/exp.s | 37 | ||||
-rw-r--r-- | src/math/i386/log1p.s | 9 | ||||
-rw-r--r-- | src/math/i386/log1pf.s | 10 |
6 files changed, 98 insertions, 3 deletions
diff --git a/src/math/i386/asin.s b/src/math/i386/asin.s index 932c7542..a9f691bf 100644 --- a/src/math/i386/asin.s +++ b/src/math/i386/asin.s @@ -2,7 +2,18 @@ .type asinf,@function asinf: flds 4(%esp) - jmp 1f + mov 4(%esp),%eax + add %eax,%eax + cmp $0x01000000,%eax + jae 1f + # subnormal x, return x with underflow + fnstsw %ax + and $16,%ax + jnz 2f + fld %st(0) + fmul %st(1) + fstps 4(%esp) +2: ret .global asinl .type asinl,@function @@ -14,6 +25,16 @@ asinl: .type asin,@function asin: fldl 4(%esp) + mov 8(%esp),%eax + add %eax,%eax + cmp $0x00200000,%eax + jae 1f + # subnormal x, return x with underflow + fnstsw %ax + and $16,%ax + jnz 2f + fsts 4(%esp) +2: ret 1: fld %st(0) fld1 fsub %st(0),%st(1) diff --git a/src/math/i386/atan.s b/src/math/i386/atan.s index 7e28b395..d73137b2 100644 --- a/src/math/i386/atan.s +++ b/src/math/i386/atan.s @@ -2,6 +2,16 @@ .type atan,@function atan: fldl 4(%esp) + mov 8(%esp),%eax + add %eax,%eax + cmp $0x00200000,%eax + jb 1f fld1 fpatan ret + # subnormal x, return x with underflow +1: fnstsw %ax + and $16,%ax + jnz 2f + fsts 4(%esp) +2: ret diff --git a/src/math/i386/atanf.s b/src/math/i386/atanf.s index 3cd40233..8caddefa 100644 --- a/src/math/i386/atanf.s +++ b/src/math/i386/atanf.s @@ -2,6 +2,18 @@ .type atanf,@function atanf: flds 4(%esp) + mov 4(%esp),%eax + add %eax,%eax + cmp $0x01000000,%eax + jb 1f fld1 fpatan ret + # subnormal x, return x with underflow +1: fnstsw %ax + and $16,%ax + jnz 2f + fld %st(0) + fmul %st(1) + fstps 4(%esp) +2: ret diff --git a/src/math/i386/exp.s b/src/math/i386/exp.s index e3b42af5..e5f54588 100644 --- a/src/math/i386/exp.s +++ b/src/math/i386/exp.s @@ -2,7 +2,18 @@ .type expm1f,@function expm1f: flds 4(%esp) - jmp 1f + mov 4(%esp),%eax + add %eax,%eax + cmp $0x01000000,%eax + jae 1f + # subnormal x, return x with underflow + fnstsw %ax + and $16,%ax + jnz 2f + fld %st(0) + fmul %st(1) + fstps 4(%esp) +2: ret .global expm1l .type expm1l,@function @@ -14,10 +25,32 @@ expm1l: .type expm1,@function expm1: fldl 4(%esp) + mov 8(%esp),%eax + add %eax,%eax + cmp $0x00200000,%eax + jae 1f + # subnormal x, return x with underflow + fnstsw %ax + and $16,%ax + jnz 2f + fsts 4(%esp) +2: ret 1: fldl2e fmulp + mov $0xc2820000,%eax + push %eax + flds (%esp) + pop %eax + fucomp %st(1) + fnstsw %ax + sahf fld1 - fld %st(1) + jb 1f + # x*log2e < -65, return -1 without underflow + fstp %st(1) + fchs + ret +1: fld %st(1) fabs fucom %st(1) fnstsw %ax diff --git a/src/math/i386/log1p.s b/src/math/i386/log1p.s index 9971e53c..6b6929c7 100644 --- a/src/math/i386/log1p.s +++ b/src/math/i386/log1p.s @@ -7,9 +7,18 @@ log1p: fldl 4(%esp) cmp $0x3fd28f00,%eax ja 1f + cmp $0x00100000,%eax + jb 2f fyl2xp1 ret 1: fld1 faddp fyl2x ret + # subnormal x, return x with underflow +2: fnstsw %ax + and $16,%ax + jnz 1f + fsts 4(%esp) + fstp %st(1) +1: ret diff --git a/src/math/i386/log1pf.s b/src/math/i386/log1pf.s index 2680a8a6..c0bcd30f 100644 --- a/src/math/i386/log1pf.s +++ b/src/math/i386/log1pf.s @@ -7,9 +7,19 @@ log1pf: flds 4(%esp) cmp $0x3e940000,%eax ja 1f + cmp $0x00800000,%eax + jb 2f fyl2xp1 ret 1: fld1 faddp fyl2x ret + # subnormal x, return x with underflow +2: fnstsw %ax + and $16,%ax + jnz 1f + fxch + fmul %st(1) + fstps 4(%esp) +1: ret |