| Submitter | Paul Clarke |
|---|---|
| Date | Nov. 21, 2019, 9:24 p.m. |
| Message ID | <1574371456-1788-1-git-send-email-pc@us.ibm.com> |
| Download | mbox | patch |
| Permalink | /patch/36104/ |
| State | New |
| Headers | show |
Comments
On Thu, 21 Nov 2019, Paul A. Clarke wrote: > From: "Paul A. Clarke" <pc@us.ibm.com> > > Very recent commit 854e91bf6b4221f424ffa13b9ef50f35623b7b74 enabled > inline of issignalingf() in general (__issignalingf in include/math.h). > There is another implementation for an inline use of issignalingf > (issignalingf_inline in sysdeps/ieee754/flt-32/math_config.h) > which could instead make use of the new enablement. > > Replace the use of issignalingf_inline with __issignaling. Using > issignaling (instead of __issignalingf) will allow future enhancements > to the type-generic implementation, issignaling, to be automatically > adopted. This version is OK.
Patch
diff --git a/sysdeps/ieee754/flt-32/e_powf.c b/sysdeps/ieee754/flt-32/e_powf.c index 4947ae2..9c1902f 100644 --- a/sysdeps/ieee754/flt-32/e_powf.c +++ b/sysdeps/ieee754/flt-32/e_powf.c @@ -158,9 +158,9 @@ __powf (float x, float y) if (__glibc_unlikely (zeroinfnan (iy))) { if (2 * iy == 0) - return issignalingf_inline (x) ? x + y : 1.0f; + return issignaling (x) ? x + y : 1.0f; if (ix == 0x3f800000) - return issignalingf_inline (y) ? x + y : 1.0f; + return issignaling (y) ? x + y : 1.0f; if (2 * ix > 2u * 0x7f800000 || 2 * iy > 2u * 0x7f800000) return x + y; if (2 * ix == 2 * 0x3f800000) diff --git a/sysdeps/ieee754/flt-32/math_config.h b/sysdeps/ieee754/flt-32/math_config.h index c5e9299..57cf441 100644 --- a/sysdeps/ieee754/flt-32/math_config.h +++ b/sysdeps/ieee754/flt-32/math_config.h @@ -101,15 +101,6 @@ asdouble (uint64_t i) return u.f; } -static inline int -issignalingf_inline (float x) -{ - uint32_t ix = asuint (x); - if (HIGH_ORDER_BIT_IS_SET_FOR_SNAN) - return (ix & 0x7fc00000) == 0x7fc00000; - return 2 * (ix ^ 0x00400000) > 2u * 0x7fc00000; -} - #define NOINLINE __attribute__ ((noinline)) attribute_hidden float __math_oflowf (uint32_t);
From: "Paul A. Clarke" <pc@us.ibm.com> Very recent commit 854e91bf6b4221f424ffa13b9ef50f35623b7b74 enabled inline of issignalingf() in general (__issignalingf in include/math.h). There is another implementation for an inline use of issignalingf (issignalingf_inline in sysdeps/ieee754/flt-32/math_config.h) which could instead make use of the new enablement. Replace the use of issignalingf_inline with __issignaling. Using issignaling (instead of __issignalingf) will allow future enhancements to the type-generic implementation, issignaling, to be automatically adopted. The implementations are slightly different, and compile to slightly different code, but I measured no significant performance difference. The second implementation was brought to my attention by: Suggested-by: Joseph Myers <joseph@codesourcery.com> --- v2: Use type-generic issignaling instead of __issignalingf, as suggested by Joseph Myers. sysdeps/ieee754/flt-32/e_powf.c | 4 ++-- sysdeps/ieee754/flt-32/math_config.h | 9 --------- 2 files changed, 2 insertions(+), 11 deletions(-)