[Bug math/5768] New: isnan() will throw SIGFPE with signalling NaN

[Bug math/5768] New: isnan() will throw SIGFPE with signalling NaN

[sjmunroe at us dot ibm dot com]

On powerpc the default implementation uses:


  savedstate = fegetenv_register ();
  reset_fpscr_bit (FPSCR_VE);
  result = !(x == x);
  fesetenv_register (savedstate);
  return result;

But with GCC-4.1+ there is no dependency to prevent GCC from moving the
fesetenv_register() [mtfsf] ahead of the (x==x) [fcmpu f1,f1]. In the default
mode this is not a problem. but if FE_INVALID is enabled and a signaling NaN is
passed to isnan() we will get a SIGFPE.

-- 
           Summary: isnan() will throw SIGFPE with signalling NaN
           Product: glibc
           Version: unspecified
            Status: NEW
          Severity: normal
          Priority: P3
         Component: math
        AssignedTo: aj at suse dot de
        ReportedBy: sjmunroe at us dot ibm dot com
                CC: glibc-bugs at sources dot redhat dot com
 GCC build triplet: powerpc64-*-linux
  GCC host triplet: powerpc64-*-linux
GCC target triplet: powerpc64-*-linux and powerpc-*-linux


http://sourceware.org/bugzilla/show_bug.cgi?id=5768

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug math/5768] isnan() will throw SIGFPE with signalling NaN

[sjmunroe at us dot ibm dot com]

------- Additional Comments From sjmunroe at us dot ibm dot com  2008-02-18 23:07 -------
Created an attachment (id=2270)
 --> (http://sourceware.org/bugzilla/attachment.cgi?id=2270&action=view)
Test case for signalling NaNs

Compile with:

gcc -O2 -g -std=gnu99 -fno-builtin test-isnan.c -lm -o test-isnan

-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=5768

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug math/5768] isnan() will throw SIGFPE with signalling NaN

[sjmunroe at us dot ibm dot com]

------- Additional Comments From sjmunroe at us dot ibm dot com  2008-02-18 23:12 -------
For powerpc[64] isnan() and isnanf() fail since they share the same
implementation. isnanl() is Ok because it uses the generic implementation from
/ieee754/ldbl-128ibm/. isinf, isfinite, isnormal, and fpclassify are also OK.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |ASSIGNED


http://sourceware.org/bugzilla/show_bug.cgi?id=5768

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug math/5768] isnan() will throw SIGFPE with signalling NaN

[sjmunroe at us dot ibm dot com]

------- Additional Comments From sjmunroe at us dot ibm dot com  2008-02-18 23:20 -------
Created an attachment (id=2271)
 --> (http://sourceware.org/bugzilla/attachment.cgi?id=2271&action=view)
Patch to replace the failing isnan.c with correct isnan.S implementation

This is simple enough to write in assembler and prevent GCC from rearranging
the code.

-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=5768

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug math/5768] isnan() will throw SIGFPE with signalling NaN

[sjmunroe at us dot ibm dot com]

------- Additional Comments From sjmunroe at us dot ibm dot com  2008-02-22 21:37 -------
Created an attachment (id=2289)
 --> (http://sourceware.org/bugzilla/attachment.cgi?id=2289&action=view)
Add Signalling NaN test to make check

This patch adds a signaling NaN test to make check

-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=5768

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug math/5768] isnan() will throw SIGFPE with signalling NaN

[drepper at redhat dot com]

------- Additional Comments From drepper at redhat dot com  2008-04-10 14:18 -------
Changes applied to cvs.

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


http://sourceware.org/bugzilla/show_bug.cgi?id=5768

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug math/5768] isnan() will throw SIGFPE with signalling NaN

[fweimer at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=5768

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.