* [PATCH] Fix error in fdim/f for infinities
@ 2020-03-10 10:24 Fabian Schriever
2020-03-10 14:47 ` Corinna Vinschen
0 siblings, 1 reply; 2+ messages in thread
From: Fabian Schriever @ 2020-03-10 10:24 UTC (permalink / raw)
To: newlib; +Cc: Fabian Schriever
The comparison c == FP_INFINITE causes the function to return +inf as it
expects x = +inf to always be larger than y. This shortcut causes
several issues as it also returns +inf for the following cases:
- fdim(+inf, +inf), expected (as per C99): +0.0
- fdim(-inf, any non NaN), expected: +0.0
I don't see a reason to keep the comparison as all the infinity cases
return the correct result using just the ternary operation.
---
newlib/libm/common/s_fdim.c | 5 +----
newlib/libm/common/sf_fdim.c | 5 +----
2 files changed, 2 insertions(+), 8 deletions(-)
diff --git a/newlib/libm/common/s_fdim.c b/newlib/libm/common/s_fdim.c
index 73a027953..61a4908f3 100644
--- a/newlib/libm/common/s_fdim.c
+++ b/newlib/libm/common/s_fdim.c
@@ -49,11 +49,8 @@ ANSI C, POSIX.
double y;
#endif
{
- int c = __fpclassifyd(x);
- if (c == FP_NAN) return(x);
+ if (__fpclassifyd(x) == FP_NAN) return(x);
if (__fpclassifyd(y) == FP_NAN) return(y);
- if (c == FP_INFINITE)
- return HUGE_VAL;
return x > y ? x - y : 0.0;
}
diff --git a/newlib/libm/common/sf_fdim.c b/newlib/libm/common/sf_fdim.c
index fe349098b..8fee57002 100644
--- a/newlib/libm/common/sf_fdim.c
+++ b/newlib/libm/common/sf_fdim.c
@@ -14,11 +14,8 @@
float y;
#endif
{
- int c = __fpclassifyf(x);
- if (c == FP_NAN) return(x);
+ if (__fpclassifyf(x) == FP_NAN) return(x);
if (__fpclassifyf(y) == FP_NAN) return(y);
- if (c == FP_INFINITE)
- return HUGE_VALF;
return x > y ? x - y : 0.0;
}
--
2.24.1.windows.2
^ permalink raw reply [flat|nested] 2+ messages in thread
* Re: [PATCH] Fix error in fdim/f for infinities
2020-03-10 10:24 [PATCH] Fix error in fdim/f for infinities Fabian Schriever
@ 2020-03-10 14:47 ` Corinna Vinschen
0 siblings, 0 replies; 2+ messages in thread
From: Corinna Vinschen @ 2020-03-10 14:47 UTC (permalink / raw)
To: Fabian Schriever; +Cc: newlib
[-- Attachment #1: Type: text/plain, Size: 1784 bytes --]
On Mar 10 11:24, Fabian Schriever wrote:
> The comparison c == FP_INFINITE causes the function to return +inf as it
> expects x = +inf to always be larger than y. This shortcut causes
> several issues as it also returns +inf for the following cases:
> - fdim(+inf, +inf), expected (as per C99): +0.0
> - fdim(-inf, any non NaN), expected: +0.0
>
> I don't see a reason to keep the comparison as all the infinity cases
> return the correct result using just the ternary operation.
> ---
> newlib/libm/common/s_fdim.c | 5 +----
> newlib/libm/common/sf_fdim.c | 5 +----
> 2 files changed, 2 insertions(+), 8 deletions(-)
>
> diff --git a/newlib/libm/common/s_fdim.c b/newlib/libm/common/s_fdim.c
> index 73a027953..61a4908f3 100644
> --- a/newlib/libm/common/s_fdim.c
> +++ b/newlib/libm/common/s_fdim.c
> @@ -49,11 +49,8 @@ ANSI C, POSIX.
> double y;
> #endif
> {
> - int c = __fpclassifyd(x);
> - if (c == FP_NAN) return(x);
> + if (__fpclassifyd(x) == FP_NAN) return(x);
> if (__fpclassifyd(y) == FP_NAN) return(y);
> - if (c == FP_INFINITE)
> - return HUGE_VAL;
>
> return x > y ? x - y : 0.0;
> }
> diff --git a/newlib/libm/common/sf_fdim.c b/newlib/libm/common/sf_fdim.c
> index fe349098b..8fee57002 100644
> --- a/newlib/libm/common/sf_fdim.c
> +++ b/newlib/libm/common/sf_fdim.c
> @@ -14,11 +14,8 @@
> float y;
> #endif
> {
> - int c = __fpclassifyf(x);
> - if (c == FP_NAN) return(x);
> + if (__fpclassifyf(x) == FP_NAN) return(x);
> if (__fpclassifyf(y) == FP_NAN) return(y);
> - if (c == FP_INFINITE)
> - return HUGE_VALF;
>
> return x > y ? x - y : 0.0;
> }
> --
> 2.24.1.windows.2
>
Pushed.
Thanks,
Corinna
--
Corinna Vinschen
Cygwin Maintainer
Red Hat
[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2020-03-10 14:47 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-03-10 10:24 [PATCH] Fix error in fdim/f for infinities Fabian Schriever
2020-03-10 14:47 ` Corinna Vinschen
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).