[Bug math/14343] New: Function logb() in math.h produces incorrect results for small inputs

[Bug math/14343] New: Function logb() in math.h produces incorrect results for small inputs

[akessler at mit dot edu]

http://sourceware.org/bugzilla/show_bug.cgi?id=14343

             Bug #: 14343
           Summary: Function logb() in math.h produces incorrect results
                    for small inputs
           Product: glibc
           Version: 2.15
            Status: NEW
          Severity: critical
          Priority: P2
         Component: math
        AssignedTo: unassigned@sourceware.org
        ReportedBy: akessler@mit.edu
    Classification: Unclassified


The function double logb (double x) produces results that are wildly incorrect
for values of x with |x| < 1 (equivalently, when logb(x) should return a
negative result). If x is written as MANTISSA * 10^EXP, logb is expected to
return EXP. Instead, for inputs with EXP < 0, logb seems to return the value
2^32 + EXP.

This is very easily reproduced on all 64-bit systems I have tried (certainly
for Linux 3.4.4, and earlier versions too) with GLIBC 2.15. I do not know if
2.16 or later have this issue, but versions prior to GLIBC 2.15 (certainly 2.13
or earlier) do not have this issue.

Steps to reproduce:

1. Compile the following program with "gcc -otest test.c -lm"
2. Run

#include <math.h>
#include <stdio.h>

int main ()
{
        int i;
        double x = 0.5;
        // Following two lines should be the same!
        printf ("%lf\n", logb (0.5));
        printf ("%lf\n", logb (x));

        return 0;
}

Actual output:
-1.000000
4294967295.000000

Expected output:
-1.000000
-1.000000

It is clear that something about the logb() function from math.h is at fault
here. The reason that the first line of this output is correct seems to be
compiler optimization -- GCC optimizes out the logb() call for the constant 0.5
and replaces it with -1, but the other call to logb is left as-is.

-- 
Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

[Bug math/14343] Function logb() in math.h produces incorrect results for small inputs

[akessler at mit dot edu]

http://sourceware.org/bugzilla/show_bug.cgi?id=14343

Andre Kessler <akessler at mit dot edu> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |akessler at mit dot edu

-- 
Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

[Bug math/14343] Function logb() in math.h produces incorrect results for small inputs

[philomath868 at gmail dot com]

http://sourceware.org/bugzilla/show_bug.cgi?id=14343

philomath868 at gmail dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |philomath868 at gmail dot
                   |                            |com

--- Comment #1 from philomath868 at gmail dot com 2012-07-10 08:38:43 UTC ---
Works correctly on glibc-2.16, x86_64 GNU/Linux 3.4.4
(releated to #13954 ?).

-- 
Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

[Bug math/14343] Function logb() in math.h produces incorrect results for small inputs

[philomath868 at gmail dot com]

http://sourceware.org/bugzilla/show_bug.cgi?id=14343

philomath868 at gmail dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|philomath868 at gmail dot   |
                   |com                         |

-- 
Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

[Bug math/14343] Function logb() in math.h produces incorrect results for small inputs

[carlos_odonell at mentor dot com]

http://sourceware.org/bugzilla/show_bug.cgi?id=14343

Carlos O'Donell <carlos_odonell at mentor dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |WAITING
   Last reconfirmed|                            |2012-07-23
                 CC|                            |carlos_odonell at mentor
                   |                            |dot com

--- Comment #2 from Carlos O'Donell <carlos_odonell at mentor dot com> 2012-07-24 02:30:18 UTC ---
Do you need a fix for this on the 2.15 stable branch?

Is it enough that 2.16 has the fix?

If you are compiling glibc yourself you can upgrade to 2.16.

If you use your distro provided glibc then you'll eventually ugprade to 2.16
and get the fix.

Confirmed that 2.15 has this bug.

carlos@domU-12-31-39-06-F5-7B:~$ ./test1
-1.000000
4294967295.000000

Confirmed that 2.16 does not have this bug.

carlos@domU-12-31-39-06-F5-7B:~/build/glibc-2.16$ ./elf/ld.so --library-path
/home/carlos/build/glibc-2.16/:/home/carlos/build/glibc-2.16/math
/home/carlos/test1
-1.000000
-1.000000

-- 
Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

[Bug math/14343] Function logb() in math.h produces incorrect results for small inputs

[akessler at mit dot edu]

http://sourceware.org/bugzilla/show_bug.cgi?id=14343

--- Comment #3 from Andre Kessler <akessler at mit dot edu> 2012-07-24 08:10:56 UTC ---
(In reply to comment #2)
> Do you need a fix for this on the 2.15 stable branch?
> 
> Is it enough that 2.16 has the fix?
> 
> If you are compiling glibc yourself you can upgrade to 2.16.

For some computers I use, it's fine that 2.16 has the fix, but I also have to
deal with some enterprise machines that will be using glibc 2.15 for quite a
while (the bug was discovered while testing some internal math libraries on
these machines). Since this problem is quite widespread -- anyone with 64-bit
glibc 2.15 will have an issue -- and this affects basic math functionality,
there really should be a fix on the 2.15 stable branch...

-- 
Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

[Bug math/14343] Function logb() in math.h produces incorrect results for small inputs

[carlos_odonell at mentor dot com]

http://sourceware.org/bugzilla/show_bug.cgi?id=14343

Carlos O'Donell <carlos_odonell at mentor dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|WAITING                     |SUSPENDED

--- Comment #4 from Carlos O'Donell <carlos_odonell at mentor dot com> 2012-07-24 13:32:11 UTC ---
(In reply to comment #3)
> (In reply to comment #2)
> > Do you need a fix for this on the 2.15 stable branch?
> > 
> > Is it enough that 2.16 has the fix?
> > 
> > If you are compiling glibc yourself you can upgrade to 2.16.
> 
> For some computers I use, it's fine that 2.16 has the fix, but I also have to
> deal with some enterprise machines that will be using glibc 2.15 for quite a
> while (the bug was discovered while testing some internal math libraries on
> these machines). Since this problem is quite widespread -- anyone with 64-bit
> glibc 2.15 will have an issue -- and this affects basic math functionality,
> there really should be a fix on the 2.15 stable branch...

Andre,

Thanks for the response. If you would like to see progress made on this I would
suggest the following course of action:
* Checkout glibc 2.15
* Apply the change made for BZ #13954 e.g.
http://sourceware.org/git/?p=glibc.git;a=commit;h=89c9aa491a7cee97bf78a29cddbf0a25c902a671
* Build glibc and run the testsuite.
* Build your application against the newly built glibc using -Wl,-rpath and
-Wl,--dynamic-linker and see if it fixes your issue.

If the newly built glibc 2.15 fixes your issue then do the following:
* Add the keyword glibc_2.15 to this issue indicating that you are requesting a
backport to the 2.15 stable branch.
* Mention the exact commit that you backported and tested, and include your
testsuite results and the machine you tested on.

At that point the 2.15 stable release branch manager can *easily* pickup the
fix for inclusion in the next 2.15 stable release e.g. 2.15.1.

I'm marking this bug as suspended pending help from someone to work on
backporting and testing a fix.

If you have *any* questions about how to do any of the above steps we would be
more than welcome to walk you through it on libc-help@sourceware.org. Please
don't hesitate to ask, no question is wrong :-)

-- 
Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

[Bug math/14343] Function logb() in math.h produces incorrect results for small inputs

[carlos at systemhalted dot org]

http://sourceware.org/bugzilla/show_bug.cgi?id=14343

Carlos O'Donell <carlos at systemhalted dot org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|carlos_odonell at mentor    |carlos at systemhalted dot
                   |dot com                     |org

-- 
Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

[Bug math/14343] [2.15] Function logb() in math.h produces incorrect results for small inputs

[jsm28 at gcc dot gnu.org]

http://sourceware.org/bugzilla/show_bug.cgi?id=14343

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|Function logb() in math.h   |[2.15] Function logb() in
                   |produces incorrect results  |math.h produces incorrect
                   |for small inputs            |results for small inputs

-- 
Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

[Bug math/14343] [2.15] Function logb() in math.h produces incorrect results for small inputs

[carlos at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=14343

Carlos O'Donell <carlos at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|SUSPENDED                   |RESOLVED
         Resolution|---                         |WONTFIX

--- Comment #5 from Carlos O'Donell <carlos at redhat dot com> ---
Needs backported patch. Reopen once you have a backported patch ready.

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

[Bug math/14343] [2.15] Function logb() in math.h produces incorrect results for small inputs

[fweimer at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=14343

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.