public inbox for gcc-bugs@sourceware.org
help / color / mirror / Atom feed
* [Bug c/96793] New: __builtin_floor produces wrong result when rounding direction is FE_DOWNWARD
@ 2020-08-26  9:10 chfast at gmail dot com
  2020-08-26  9:11 ` [Bug c/96793] " chfast at gmail dot com
                   ` (23 more replies)
  0 siblings, 24 replies; 25+ messages in thread
From: chfast at gmail dot com @ 2020-08-26  9:10 UTC (permalink / raw)
  To: gcc-bugs

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=96793

            Bug ID: 96793
           Summary: __builtin_floor produces wrong result when rounding
                    direction is FE_DOWNWARD
           Product: gcc
           Version: 10.2.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c
          Assignee: unassigned at gcc dot gnu.org
          Reporter: chfast at gmail dot com
  Target Milestone: ---

Created attachment 49125
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=49125&action=edit
Preprocessed test code

If the rounding direction is set to FE_DOWNWARD by fesetround(),
the __builtin_floor() result is -0 where it should be +0. Eg.
__builtin_floor(0.25) == -0.

The outputs are done with GCC 10.2.0 (x86_64-linux-gnu) with -O2
-frounding-math -lm. Full gcc -v output in cli.log file. This is also
reproducible in GCC 9 and trunk.


The test code to reproduce the bug checks if the result's sign is 0 as expected
(attached as builtin_floor_test.c):

enum { FE_DOWNWARD = 0x400 };

extern int fesetround(int rounding_direction);

__attribute__((noinline))
float builtin_floorf(float value)
{
    return __builtin_floorf(value);
}

int main()
{
    fesetround(FE_DOWNWARD);
    float result = builtin_floorf(0.25f);
    return __builtin_signbitf(result) != 0;
}


The __builtin_floor() generates the following assembly (part of
builtin_floor_test.s):

        movss   .LC1(%rip), %xmm2
        movss   .LC0(%rip), %xmm4
        movaps  %xmm0, %xmm3
        movaps  %xmm0, %xmm1
        andps   %xmm2, %xmm3
        ucomiss %xmm3, %xmm4
        jbe     .L2
        cvttss2sil      %xmm0, %eax
        pxor    %xmm3, %xmm3
        andnps  %xmm1, %xmm2
        cvtsi2ssl       %eax, %xmm3
        movaps  %xmm3, %xmm4
        cmpnless        %xmm0, %xmm4
        movss   .LC2(%rip), %xmm0
        andps   %xmm0, %xmm4
        subss   %xmm4, %xmm3
        movaps  %xmm3, %xmm0
        orps    %xmm2, %xmm0
.L2:
        ret

^ permalink raw reply	[flat|nested] 25+ messages in thread
end of thread, other threads:[~2020-12-23  8:21 UTC | newest]

Thread overview: 25+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-08-26  9:10 [Bug c/96793] New: __builtin_floor produces wrong result when rounding direction is FE_DOWNWARD chfast at gmail dot com
2020-08-26  9:11 ` [Bug c/96793] " chfast at gmail dot com
2020-08-26  9:11 ` chfast at gmail dot com
2020-08-26  9:12 ` chfast at gmail dot com
2020-08-26  9:21 ` chfast at gmail dot com
2020-08-26 10:46 ` [Bug target/96793] " rguenth at gcc dot gnu.org
2020-08-26 11:17 ` ubizjak at gmail dot com
2020-08-28  7:26 ` ubizjak at gmail dot com
2020-08-28  7:26 ` ubizjak at gmail dot com
2020-08-28  8:58 ` chfast at gmail dot com
2020-08-28  9:00 ` ubizjak at gmail dot com
2020-08-28  9:40 ` rguenth at gcc dot gnu.org
2020-08-28  9:48 ` ubizjak at gmail dot com
2020-08-28  9:54 ` ubizjak at gmail dot com
2020-08-28 11:31 ` glisse at gcc dot gnu.org
2020-08-28 12:20 ` glisse at gcc dot gnu.org
2020-08-28 14:06 ` chfast at gmail dot com
2020-08-28 14:31 ` chfast at gmail dot com
2020-08-28 14:37 ` ubizjak at gmail dot com
2020-12-22 12:17 ` ubizjak at gmail dot com
2020-12-22 17:20 ` ubizjak at gmail dot com
2020-12-22 20:13 ` ubizjak at gmail dot com
2020-12-23  8:20 ` ubizjak at gmail dot com
2020-12-23  8:21 ` ubizjak at gmail dot com
2020-12-23  8:21 ` ubizjak at gmail dot com

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).