[Bug math/16339] New: log1p missing underflows

[Bug math/16339] New: log1p missing underflows

[jsm28 at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=16339

            Bug ID: 16339
           Summary: log1p missing underflows
           Product: glibc
           Version: 2.19
            Status: NEW
          Severity: normal
          Priority: P2
         Component: math
          Assignee: unassigned at sourceware dot org
          Reporter: jsm28 at gcc dot gnu.org

log1p (subnormal) produces an inexact result so should raise the underflow
exception.  But on x86_64, log1pf (0x1p-149) and log1p (0x1p-1074) do not raise
that exception,

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug math/16339] log1p missing underflows

[zimmerma+gcc at loria dot fr]

https://sourceware.org/bugzilla/show_bug.cgi?id=16339

Paul Zimmermann <zimmerma+gcc at loria dot fr> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |zimmerma+gcc at loria dot fr

--- Comment #1 from Paul Zimmermann <zimmerma+gcc at loria dot fr> ---
on the other hand we get some spurious underflow exceptions, for example for
long double on x86_64:

Testing function log1p for exponent 16383 [seed=8473].
 rounding mode MPFR_RNDD:
      wrong underflow flag: mpfr gives 0, library 16
      x=-3.36210314311209350626e-4932
      library gives -3.36210314311209350626e-4932
      mpfr    gives -3.36210314311209350663e-4932

The glibc result is incorrect since log(1+x) = x - x^2/2 + o(x^2), thus it is
smaller than x for tiny x, and rounding down should give nextbelow(x).

Since |x| is the smallest normal number, no underflow exception should be
raised.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug math/16339] log1p missing underflows

[joseph at codesourcery dot com]

http://sourceware.org/bugzilla/show_bug.cgi?id=16339

--- Comment #2 from joseph at codesourcery dot com <joseph at codesourcery dot com> ---
Does not seem spurious (consistent with a slightly inaccurate 
infinite-precision value).  In any case, spurious exceptions are almost 
always separate bugs from missing exceptions.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug math/16339] log1p missing underflows

[zimmerma+gcc at loria dot fr]

https://sourceware.org/bugzilla/show_bug.cgi?id=16339

--- Comment #3 from Paul Zimmermann <zimmerma+gcc at loria dot fr> ---
(In reply to joseph@codesourcery.com from comment #2)
> Does not seem spurious (consistent with a slightly inaccurate 
> infinite-precision value).  In any case, spurious exceptions are almost 
> always separate bugs from missing exceptions.

ok, I retract Comment 1.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug math/16339] log1p missing underflows

[fweimer at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=16339

Florian Weimer <fweimer at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
              Flags|                            |security-

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug math/16339] log1p missing underflows

[jsm28 at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=16339

Joseph Myers <jsm28 at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |nszabolcs at gmail dot com

--- Comment #4 from Joseph Myers <jsm28 at gcc dot gnu.org> ---
*** Bug 18248 has been marked as a duplicate of this bug. ***

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug math/16339] log1p missing underflows

[jsm28 at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=16339

Joseph Myers <jsm28 at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|---                         |FIXED

--- Comment #6 from Joseph Myers <jsm28 at gcc dot gnu.org> ---
Fixed for 2.22.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug math/16339] log1p missing underflows

[cvs-commit at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=16339

--- Comment #5 from cvs-commit at gcc dot gnu.org <cvs-commit at gcc dot gnu.org> ---
This is an automated email from the git hooks/post-receive script. It was
generated because a ref change was pushed to the repository containing
the project "GNU C Library master sources".

The branch, master has been updated
       via  0b7a5f920163d03806d7c5d9d1c83b16942c9496 (commit)
      from  95b07fbcc702a83b421cb5d0488ba2817d3123c1 (commit)

Those revisions listed above that are new to this repository have
not appeared on any other notification email; so we list those
revisions in full, below.

- Log -----------------------------------------------------------------
https://sourceware.org/git/gitweb.cgi?p=glibc.git;h=0b7a5f920163d03806d7c5d9d1c83b16942c9496

commit 0b7a5f920163d03806d7c5d9d1c83b16942c9496
Author: Joseph Myers <joseph@codesourcery.com>
Date:   Thu May 14 23:38:07 2015 +0000

    Fix log1p missing underflows (bug 16339).

    Similar to various other bugs in this area, some log1p implementations
    do not raise the underflow exception for subnormal arguments, when the
    result is tiny and inexact.  This patch forces the exception in a
    similar way to previous fixes.  (The ldbl-128ibm implementation
    doesn't currently need any change as it already generates this
    exception, albeit through code that would generate spurious exceptions
    in other cases; special code for this issue will only be needed there
    when fixing the spurious exceptions.)

    Tested for x86_64, x86, powerpc and mips64.

        [BZ #16339]
        * sysdeps/i386/fpu/s_log1p.S (dbl_min): New object.
        (__log1p): Force underflow exception for results with small
        absolute value.
        * sysdeps/i386/fpu/s_log1pf.S (flt_min): New object.
        (__log1pf): Force underflow exception for results with small
        absolute value.
        * sysdeps/ieee754/dbl-64/s_log1p.c: Include <float.h>.
        (__log1p): Force underflow exception for results with small
        absolute value.
        * sysdeps/ieee754/flt-32/s_log1pf.c: Include <float.h>.
        (__log1pf): Force underflow exception for results with small
        absolute value.
        * sysdeps/ieee754/ldbl-128/s_log1pl.c: Include <float.h>.
        (__log1pl): Force underflow exception for results with small
        absolute value.
        * math/auto-libm-test-in: Do not allow missing underflow
        exceptions from log1p.
        * math/auto-libm-test-out: Regenerated.

-----------------------------------------------------------------------

Summary of changes:
 ChangeLog                           |   22 ++++
 NEWS                                |    2 +-
 math/auto-libm-test-in              |    9 +-
 math/auto-libm-test-out             |  200 +++++++++++++++++-----------------
 sysdeps/i386/fpu/s_log1p.S          |   27 +++++-
 sysdeps/i386/fpu/s_log1pf.S         |   27 +++++-
 sysdeps/ieee754/dbl-64/s_log1p.c    |   10 ++-
 sysdeps/ieee754/flt-32/s_log1pf.c   |    8 ++
 sysdeps/ieee754/ldbl-128/s_log1pl.c |    6 +
 9 files changed, 202 insertions(+), 109 deletions(-)

-- 
You are receiving this mail because:
You are on the CC list for the bug.