unpack(NaN) big baddaboom

[p5pRT]

Migrated from rt.perl.org#2269 (status was 'resolved')

Searchable as RT2269$

[p5pRT]

From iang@systemics.com

Created by iang@systemics.com

Running following program causes "Floating point exception" on
FreeBSD 3.2-STABLE *and* 4.0-CURRENT #0​: Tue Feb 29 02​:11​:52 AST 2000
(but not on Linux "2.3.44 #12 SMP") (all 5.00503)

===8<======8<======8<======8<====
#!/usr/bin/perl

my $packed = "\0\0\xc0\x7f";
print STDERR "len​: ", length($packed), " bytes​: ", unpack("H*", $packed), "\n";
my $float = unpack("f", $packed);
print STDERR "float done\n";
print STDERR "float​: $float\n";

exit 0;
===8<======8<======8<======8<====
hayek$ ./moo.pl
len​: 4 bytes​: 0000c07f
float done
Floating point exception
hayek$ echo $?
136
hayek$
===8<======8<======8<======8<====

Perl Info

This perlbug was built using Perl 5.00502 - Mon Oct 26 16:01:51 AST 1998
It is being executed now by  Perl 5.00503 - $Date: 1999/05/05 19:42:40 $.

Site configuration information for perl 5.00503:

Configured by markm at $Date: 1999/05/05 19:42:40 $.

Summary of my perl5 (5.0 patchlevel 5 subversion 3) configuration:
  Platform:
    osname=freebsd, osvers=4.0-current, archname=i386-freebsd
    uname='freebsd freefall.freebsd.org 4.0-current freebsd 4.0-current #0: $Date: 1999/05/05 19:42:40 $'
    hint=recommended, useposix=true, d_sigaction=define
    usethreads=undef useperlio=undef d_sfio=undef
  Compiler:
    cc='cc', optimize='undef', gccversion=egcs-2.91.66 19990314 (egcs-1.1.2 release)
    cppflags=''
    ccflags =''
    stdchar='char', d_stdstdio=undef, usevfork=true
    intsize=4, longsize=4, ptrsize=4, doublesize=8
    d_longlong=define, longlongsize=8, d_longdbl=define, longdblsize=12
    alignbytes=4, usemymalloc=n, prototype=define
  Linker and Libraries:
    ld='cc', ldflags ='-Wl,-E'
    libpth=/usr/lib
    libs=-lm -lc -lcrypt
    libc=/usr/lib/libc.so, so=so, useshrplib=true, libperl=libperl.so.3
  Dynamic Linking:
    dlsrc=dl_dlopen.xs, dlext=so, d_dlsymun=undef, ccdlflags=' '
    cccdlflags='-DPIC -fpic', lddlflags='-shared'

Locally applied patches:
    


@INC for perl 5.00503:
    /data/ricardo/perl
    /home/iang/src/hayek/ricardo/blib/lib
    /usr/libdata/perl/5.00503/mach
    /usr/libdata/perl/5.00503
    /usr/local/lib/perl5/site_perl/5.005/i386-freebsd
    /usr/local/lib/perl5/site_perl/5.005
    .


Environment for perl 5.00503:
    HOME=/home/iang
    LANG (unset)
    LD_LIBRARY_PATH (unset)
    LOGDIR (unset)
    PATH=/home/iang/src/hayek/market/cmd:/home/iang/src/hayek/market/bin:/data/ricardo/perl/cmd:/home/iang/src/hayek/ricardo/cmd:/home/iang/bin:/home/iang/cmd:/usr/local/bin:/bin:/usr/bin:/usr/X11R6/bin:/usr/local/java/bin:/sbin:/usr/sbin:/usr/local/sbin
    PERLLIB=/data/ricardo/perl:/home/iang/src/hayek/ricardo/blib/lib
    PERL_BADLANG (unset)
    SHELL=/usr/local/bin/bash

[p5pRT]

From [Unknown Contact. See original ticket]

That is because FreeBSD correctly traps operations on the value NaN.

Depending on your religion the bug is either in perl itself which
should be more careful about NaN and Inf values, or in the perl
script you supply which does something which is patently bogus,
but nothing which should provoke a FP trap according to IEEE.

I would argue that as long as your perl script doesn't use the NaN
in an operation which IEEE documents as giving a trap ( compare of
two floats for ordered magnitude for instance) then your perl
interpreter shouldn't either.

Poul-Henning

In message <200003061831.OAA74965@​systemics.com>, Ian Grigg writes​:

This is a bug report for perl from iang@​systemics.com,
generated with the help of perlbug 1.26 running under perl 5.00503.

-----------------------------------------------------------------
[Please enter your report here]

Running following program causes "Floating point exception" on
FreeBSD 3.2-STABLE *and* 4.0-CURRENT #0​: Tue Feb 29 02​:11​:52 AST 2000
(but not on Linux "2.3.44 #12 SMP") (all 5.00503)

===8<======8<======8<======8<====
#!/usr/bin/perl

my $packed = "\0\0\xc0\x7f";
print STDERR "len​: ", length($packed), " bytes​: ", unpack("H*", $packed), "\n";
my $float = unpack("f", $packed);
print STDERR "float done\n";
print STDERR "float​: $float\n";

exit 0;
===8<======8<======8<======8<====
hayek$ ./moo.pl
len​: 4 bytes​: 0000c07f
float done
Floating point exception
hayek$ echo $?
136
hayek$
===8<======8<======8<======8<====

[Please do not change anything below this line]
-----------------------------------------------------------------

---
This perlbug was built using Perl 5.00502 - Mon Oct 26 16​:01​:51 AST 1998
It is being executed now by Perl 5.00503 - $Date​: 1999/05/05 19​:42​:40 $.

Site configuration information for perl 5.00503​:

Configured by markm at $Date​: 1999/05/05 19​:42​:40 $.

Summary of my perl5 (5.0 patchlevel 5 subversion 3) configuration​:
Platform​:
osname=freebsd, osvers=4.0-current, archname=i386-freebsd
uname='freebsd freefall.freebsd.org 4.0-current freebsd 4.0-current #0​: $Date​: 1999/05/05 19​:42​:40 $'
hint=recommended, useposix=true, d_sigaction=define
usethreads=undef useperlio=undef d_sfio=undef
Compiler​:
cc='cc', optimize='undef', gccversion=egcs-2.91.66 19990314 (egcs-1.1.2 release)
cppflags=''
ccflags =''
stdchar='char', d_stdstdio=undef, usevfork=true
intsize=4, longsize=4, ptrsize=4, doublesize=8
d_longlong=define, longlongsize=8, d_longdbl=define, longdblsize=12
alignbytes=4, usemymalloc=n, prototype=define
Linker and Libraries​:
ld='cc', ldflags ='-Wl,-E'
libpth=/usr/lib
libs=-lm -lc -lcrypt
libc=/usr/lib/libc.so, so=so, useshrplib=true, libperl=libperl.so.3
Dynamic Linking​:
dlsrc=dl_dlopen.xs, dlext=so, d_dlsymun=undef, ccdlflags=' '
cccdlflags='-DPIC -fpic', lddlflags='-shared'

Locally applied patches​:

---
@​INC for perl 5.00503​:
/data/ricardo/perl
/home/iang/src/hayek/ricardo/blib/lib
/usr/libdata/perl/5.00503/mach
/usr/libdata/perl/5.00503
/usr/local/lib/perl5/site_perl/5.005/i386-freebsd
/usr/local/lib/perl5/site_perl/5.005
.

---
Environment for perl 5.00503​:
HOME=/home/iang
LANG (unset)
LD_LIBRARY_PATH (unset)
LOGDIR (unset)
PATH=/home/iang/src/hayek/market/cmd​:/home/iang/src/hayek/market/bin​:/data/ricardo/perl/cmd​:/home/iang/src/hayek/ricardo/cmd​:/home/iang/bin​:/home/iang/cmd​:/usr/local/bin​:/bin​:/usr/bin​:/usr/X11R6/bin​:/usr/local/java/bin​:/sbin​:/usr/sbin​:/usr/local/sbin
PERLLIB=/data/ricardo/perl​:/home/iang/src/hayek/ricardo/blib/lib
PERL_BADLANG (unset)
SHELL=/usr/local/bin/bash

To Unsubscribe​: send mail to majordomo@​FreeBSD.org
with "unsubscribe freebsd-current" in the body of the message

--
Poul-Henning Kamp FreeBSD coreteam member
phk@​FreeBSD.ORG "Real hackers run -current on their laptop."
FreeBSD -- It will take a long time before progress goes too far!

[p5pRT]

From [Unknown Contact. See original ticket]

More on that!

Lads here took this to heart and put the debugger over the code.

Apparently, Linux uses glibc which sets the flags in FPU to ignore
the cmp-NaN fault. Then, I guess, Perl picks it up and turns it into
NaN. OTOH, FreeBSD libraries don't do that, and the FPU causes
the cmp-NaN fault to cause a FP exception.

(Recorded 2nd hand, you won't catch me using debuggers.)

That is because FreeBSD correctly traps operations on the value NaN.

Depending on your religion the bug is either in perl itself which
should be more careful about NaN and Inf values, or in the perl
script you supply which does something which is patently bogus,
but nothing which should provoke a FP trap according to IEEE.

I don't know much about religion, but this feels like
a Perl bug to me. I feed a highlevel command a number
I read off the wire, and it blows up. As I have no
control over the wire (it's a network) I cannot *not*
handle the number.

I would argue that as long as your perl script doesn't use the NaN
in an operation which IEEE documents as giving a trap ( compare of
two floats for ordered magnitude for instance) then your perl
interpreter shouldn't either.

Well, in that sense, this is my workaround​:

  my $data = $is->read(4);
  #
  # Damn, FreeBSD throws Floating Point Exception on NaN!
  #
  my $long = unpack("N", $data);
  if ($long == 2143289344)
  {
  return "NaN";
  }

So I'm fine with it for now. But, as Perl is an interpretive
language, and they are supposed to, on some higher plain of
thought, handle everything thrown at them, I'd rather it just
returned "NaN" like Linux.

Still, I'm fixed! Thanks for the confirmation.

iang

[p5pRT]

From [Unknown Contact. See original ticket]

In message <200003061910.PAA75240@​systemics.com>, Ian Grigg writes​:

Depending on your religion the bug is either in perl itself which
should be more careful about NaN and Inf values, or in the perl
script you supply which does something which is patently bogus,
but nothing which should provoke a FP trap according to IEEE.

I don't know much about religion, but this feels like
a Perl bug to me. I feed a highlevel command a number
I read off the wire, and it blows up. As I have no
control over the wire (it's a network) I cannot *not*
handle the number.

I would agree with that.

I would argue that as long as your perl script doesn't use the NaN
in an operation which IEEE documents as giving a trap ( compare of
two floats for ordered magnitude for instance) then your perl
interpreter shouldn't either.

Well, in that sense, this is my workaround​:

my $data = $is->read(4);
#
# Damn, FreeBSD throws Floating Point Exception on NaN!
#
my $long = unpack("N", $data);
if ($long == 2143289344)
{
return "NaN";
}

That is not enough, NaN has multiple binary representations.
isnan(3) is your friend.

--
Poul-Henning Kamp FreeBSD coreteam member
phk@​FreeBSD.ORG "Real hackers run -current on their laptop."
FreeBSD -- It will take a long time before progress goes too far!

[p5pRT]

From [Unknown Contact. See original ticket]

On 2000-Mar-07 05​:35​:04 +1100, Ian Grigg <iang@​systemics.com> wrote​:

my $packed = "\0\0\xc0\x7f";
print STDERR "len​: ", length($packed), " bytes​: ", unpack("H*", $packed), "\n";
my $float = unpack("f", $packed);
print STDERR "float done\n";
print STDERR "float​: $float\n";

Looking at the resultant core file, perl is getting SIGFPE inside
cast_i32() (probably when it tries to cast the NaN to an int).
Unfortunately, the code for cast_i32() does not handle NaNs (though it
does handle infinity). (And it's not clear what integer value should
be associated with NaN in any case).

As to whether it should trap or not, that is less clear-cut. FreeBSD
takes the view that trying to convert a NaN into an integer is not a
sensible operation, so it signals SIGFPE (the i386 NPX gives an
invalid operand exception). Other implementations may differ.

Peter

[p5pRT]

From @floatingatoll

I don't see this explosion as of @​17821, linux kernel 2.2.18pre11-va2.1smp,
glibc 2.1.3.

./perl 2269.pl
len​: 4 bytes​: 0000c07f
float done
float​: nan
echo $?
0

[p5pRT]

From @schwern

[iang@​systemics.com - Mon Mar 06 02​:33​:55 2000]​:

Running following program causes "Floating point exception" on
FreeBSD 3.2-STABLE *and* 4.0-CURRENT #0​: Tue Feb 29 02​:11​:52 AST 2000
(but not on Linux "2.3.44 #12 SMP") (all 5.00503)

===8<======8<======8<======8<====
#!/usr/bin/perl

my $packed = "\0\0\xc0\x7f";
print STDERR "len​: ", length($packed), " bytes​: ", unpack("H*",
$packed), "\n";
my $float = unpack("f", $packed);
print STDERR "float done\n";
print STDERR "float​: $float\n";

exit 0;
===8<======8<======8<======8<====
hayek$ ./moo.pl
len​: 4 bytes​: 0000c07f
float done
Floating point exception
hayek$ echo $?
136
hayek$
===8<======8<======8<======8<====

Anyone have a FreeBSD machine handy with a recent Perl on it to try this
out?

[p5pRT]

From @smpeters

On Tue, Jul 12, 2005 at 08​:09​:07PM -0700, Michael G Schwern via RT wrote​:

[iang@​systemics.com - Mon Mar 06 02​:33​:55 2000]​:

Running following program causes "Floating point exception" on
FreeBSD 3.2-STABLE *and* 4.0-CURRENT #0​: Tue Feb 29 02​:11​:52 AST 2000
(but not on Linux "2.3.44 #12 SMP") (all 5.00503)

===8<======8<======8<======8<====
#!/usr/bin/perl

my $packed = "\0\0\xc0\x7f";
print STDERR "len​: ", length($packed), " bytes​: ", unpack("H*",
$packed), "\n";
my $float = unpack("f", $packed);
print STDERR "float done\n";
print STDERR "float​: $float\n";

exit 0;
===8<======8<======8<======8<====
hayek$ ./moo.pl
len​: 4 bytes​: 0000c07f
float done
Floating point exception
hayek$ echo $?
136
hayek$
===8<======8<======8<======8<====

Anyone have a FreeBSD machine handy with a recent Perl on it to try this
out?

Testdrive probably does.

[p5pRT]

From @schwern

On Tue, Jul 12, 2005 at 10​:17​:46PM -0500, Steve Peters wrote​:

Anyone have a FreeBSD machine handy with a recent Perl on it to try this
out?

Testdrive probably does.

They do but they're not accepting connections.

SourceForge's compile farm has one... FreeBSD 4.10-BETA running 5.5.3, great.

schwern@​x86-freebsd1​:~$ perl
my $packed = "\0\0\xc0\x7f";
print STDERR "len​: ", length($packed), " bytes​: ",
\ unpack("H*", $packed), "\n";
my $float = unpack("f", $packed);
print STDERR "float done\n";
print STDERR "float​: $float\n";

exit 0;
len​: 4 bytes​: SCALAR(0x804e054)
float done
float​: NaN

--
Michael G Schwern schwern@​pobox.com http​://www.pobox.com/~schwern
Ahh email, my old friend. Do you know that revenge is a dish that is best
served cold? And it is very cold on the Internet!

[p5pRT]

From iang@systemics.com

On Tue, Jul 12, 2005 at 10​:17​:46PM -0500, Steve Peters wrote​:

Anyone have a FreeBSD machine handy with a recent Perl on it to try this
out?

What versions of "recent" are you after?

guderian$ perl -version

This is perl, v5.8.7 built for i386-freebsd-64int
(with 1 registered patch, see perl -V for more detail)

guderian$ uname -a
FreeBSD localhost 5.4-RELEASE FreeBSD 5.4-RELEASE #0​: Sun May 8 10​:21​:06 UTC 2005 root@​harlow.cse.buffalo.edu​:/usr/obj/usr/src/sys/GENERIC i386

guderian$ perl < nan
len​: 4 bytes​: 0000c07f
float done
float​: nan
guderian$ cat nan

my $packed = "\0\0\xc0\x7f";
print STDERR "len​: ", length($packed), " bytes​: ", unpack("H*", $packed), "\n";
my $float = unpack("f", $packed);
print STDERR "float done\n";
print STDERR "float​: $float\n";

exit 0;
guderian$

=================
same result on perl 5.8.5 (no patches).

5.6.2 (none) said​:

koetjeboe$ perl < nan
len​: 4 bytes​: 0000c07f
float done
float​: NaN

which differs in the spelling of NaN.

--
Advances in Financial Cryptography, Issue 2​:
  https://www.financialcryptography.com/mt/archives/000498.html
Mark Stiegler, An Introduction to Petname Systems
Nick Szabo, Scarce Objects
Ian Grigg, Triple Entry Accounting

[p5pRT]

From @schwern

On Wed, Jul 13, 2005 at 12​:00​:21PM +0100, Ian Grigg wrote​:

What versions of "recent" are you after?

These are good. So it doesn't explode, I guess that's better? (I'm not
an expert on how floats should behave).

It would be nice if the version of Perl this bug was detected on (5.5.3
I think) could be tried on the newer FreeBSD to see if its Perl or FreeBSD
which was fixed. But not necessary.

--
Michael G Schwern schwern@​pobox.com http​://www.pobox.com/~schwern
Reality is that which, when you stop believing in it, doesn't go away.
  -- Phillip K. Dick

[p5pRT]

@smpeters - Status changed from 'open' to 'resolved'