Fcntl -w causing perl script compile failures

[p5pRT]

Migrated from rt.perl.org#77762 (status was 'resolved')

Searchable as RT77762$

[p5pRT]

From @toddr

Created by @toddr

This breaks​:
$>perl5.12.1 -w -e 'use Fcntl; print "never runs"; if(0) {my $foo = Fcntl​::F_GETFL() or die};'
Found = in conditional, should be == at -e line 1.

This does not​:
$>perl5.12.1 -w -e 'use Fcntl; print "runs"; if(0) {my $foo = Fcntl​::F_GETFL()};'
runs

The warning message happens in the BEGIN block for sure. if I take -w out, the code does not fail.

I think it's related to this comment in Fcntl.pm but I'm not sure.
# Force the constants to become inlined
BEGIN {
  XSLoader​::load 'Fcntl', $VERSION;
}

Perl Info

Flags:
   category=library
   severity=low
   module=Locale::Maketext
   Type=Patch
   PatchStatus=HasPatch

Site configuration information for perl 5.12.1:

Configured by cPanel at Mon Jun  7 11:16:28 CDT 2010.

Summary of my perl5 (revision 5 version 12 subversion 1) configuration:

 Platform:
   osname=linux, osvers=2.6.18-194.3.1.el5, archname=x86_64-linux
   uname='linux rpmb-centos-50-64bit 2.6.18-194.3.1.el5 #1 smp thu may 13 13:08:30 edt 2010 x86_64 x86_64 x86_64 gnulinux '
   config_args='-des -Darchname=x86_64-linux -Dcc=/usr/local/cpanel/bin/gcc -Dcpp=/usr/local/cpanel/bin/gcc -E -DDEBUGGING=none -Doptimize=-Os -Dusemymalloc=y -Duseshrplib=true -Duselargefiles=yes -Duseposix=true -Dhint=recommended -Duseperlio=yes -Dccflags=-I/usr/local/cpanel/include -L/usr/local/cpanel/lib64 -Wl,-rpath -Wl,/usr/local/cpanel/lib64  -Dcppflags=-I/usr/local/cpanel/include -L/usr/local/cpanel/lib64 -Dldflags=-Wl,-rpath -Wl,/usr/local/cpanel/lib64 -L/usr/local/cpanel/lib64 -Dprefix=/usr/local/cpanel -Dsiteprefix=/usr/local/cpanel -Dsitebin=/usr/local/cpanel/bin -Dsitelib=/usr/local/cpanel/lib64/perl5/site_lib -Dprivlib=/usr/local/cpanel/lib64/perl5/5.12.1 -Dotherlibdirs=/var/cpanel/perl5/lib:/usr/local/cpanel/lib64/perl5/cpanel -Dman1dir=/usr/local/cpanel/share/man/man1 -Dman3dir=/usr/local/cpanel/share/man/man3 -Dsiteman1dir=/usr/local/cpanel/share/man/man1 -Dsiteman3dir=/usr/local/cpanel/share/man/man3 -Dcf_by=cPanel -Dmyhostname=localhost -Dperladmin=root@localhost -Dcf_email=support@cpanel.net -Di_dbm=/usr/local/cpanel/include -Di_gdbm=/usr/local/cpanel/include -Di_ndbm=/usr/local/cpanel/include -Ud_dosuid -Uuserelocatableinc -Umad -Uusethreads -Uusemultiplicity -Uusesocks -Uuselongdouble -Ui_db -Aldflags=-L/usr/local/cpanel/lib64 -L/usr/lib64 -L/lib64 -lgdbm -Dlocincpth=/usr/local/cpanel/include /usr/local/include  -Acflags=-fPIC -DPIC -m64 -I/usr/local/cpanel/include -Dlibpth=/usr/local/cpanel/lib64 /usr/local/lib64 /usr/local/lib /lib64 /usr/lib64  -Duse64bitint=yes -Duse64bitall=yes'
   hint=recommended, useposix=true, d_sigaction=define
   useithreads=undef, usemultiplicity=undef
   useperlio=define, d_sfio=undef, uselargefiles=define, usesocks=undef
   use64bitint=define, use64bitall=define, uselongdouble=undef
   usemymalloc=y, bincompat5005=undef
 Compiler:
   cc='/usr/local/cpanel/bin/gcc', ccflags ='-I/usr/local/cpanel/include -L/usr/local/cpanel/lib64 -Wl,-rpath -Wl,/usr/local/cpanel/lib64 -fno-strict-aliasing -pipe -fstack-protector -I/usr/local/cpanel/include -I/usr/local/include -D_LARGEFILE_SOURCE -D_FILE_OFFSET_BITS=64',
   optimize='-Os',
   cppflags='-I/usr/local/cpanel/include -L/usr/local/cpanel/lib64 -I/usr/local/cpanel/include -L/usr/local/cpanel/lib64 -Wl,-rpath -Wl,/usr/local/cpanel/lib64 -fno-strict-aliasing -pipe -fstack-protector -I/usr/local/cpanel/include -I/usr/local/include'
   ccversion='', gccversion='4.5.0', gccosandvers=''
   intsize=4, longsize=8, ptrsize=8, doublesize=8, byteorder=12345678
   d_longlong=define, longlongsize=8, d_longdbl=define, longdblsize=16
   ivtype='long', ivsize=8, nvtype='double', nvsize=8, Off_t='off_t', lseeksize=8
   alignbytes=8, prototype=define
 Linker and Libraries:
   ld='/usr/local/cpanel/bin/gcc', ldflags ='-Wl,-rpath -Wl,/usr/local/cpanel/lib64 -L/usr/local/cpanel/lib64 -L/usr/local/cpanel/lib64 -L/usr/lib64 -L/lib64 -lgdbm -fstack-protector -L/usr/local/lib'
   libpth=/usr/local/cpanel/lib64 /usr/local/lib64 /usr/local/lib /lib64 /usr/lib64
   libs=-lnsl -lgdbm -ldb -ldl -lm -lcrypt -lutil -lc -lgdbm_compat
   perllibs=-lnsl -ldl -lm -lcrypt -lutil -lc
   libc=/lib/libc-2.5.so, so=so, useshrplib=true, libperl=libperl.so
   gnulibc_version='2.5'
 Dynamic Linking:
   dlsrc=dl_dlopen.xs, dlext=so, d_dlsymun=undef, ccdlflags='-Wl,-E -Wl,-rpath,/usr/local/cpanel/lib64/perl5/5.12.1/x86_64-linux/CORE'
   cccdlflags='-fPIC', lddlflags='-shared -Os -L/usr/local/cpanel/lib64 -L/usr/lib64 -L/lib64 -L/usr/local/lib -fstack-protector'

Locally applied patches:
   cPanel Patches


@INC for perl 5.12.1:
   /usr/local/cpanel/lib64/perl5/site_lib/x86_64-linux
   /usr/local/cpanel/lib64/perl5/site_lib
   /usr/local/cpanel/lib64/perl5/5.12.1/x86_64-linux
   /usr/local/cpanel/lib64/perl5/5.12.1
   /var/cpanel/perl5/lib
   /usr/local/cpanel/lib64/perl5/cpanel
   .


Environment for perl 5.12.1:
   HOME=/home/toddr
   LANG=en_US.UTF-8
   LANGUAGE (unset)
   LD_LIBRARY_PATH (unset)
   LOGDIR (unset)
   PATH=/usr/local/cpanel/bin:/usr/local/bin:/usr/bin:/bin:/usr/sbin:/sbin:/usr/local/bin:/usr/X11/bin:
   PERL_BADLANG (unset)
   SHELL=/bin/zsh

[p5pRT]

From @toddr

This is a 5.8.8 regression for sure

/usr/bin/perl5.8.8 -w -e 'use Fcntl; print "never runs"; if(0) {my $foo = Fcntl​::F_GETFL() or die};'
never runs

It fails in 5.10 also
$>/usr/bin/perl5.10.0 -w -e 'use Fcntl; print "never runs"; if(0) {my $foo = Fcntl​::F_GETFL() or
die};'
Found = in conditional, should be == at -e line 1.

[p5pRT]

The RT System itself - Status changed from 'new' to 'open'

[p5pRT]

From @nwc10

On Thu, Sep 09, 2010 at 12​:15​:24PM -0700, Todd Rinaldo wrote​:

This breaks​:
$>perl5.12.1 -w -e 'use Fcntl; print "never runs"; if(0) {my $foo = Fcntl​::F_GETFL() or die};'
Found = in conditional, should be == at -e line 1.

This does not​:
$>perl5.12.1 -w -e 'use Fcntl; print "runs"; if(0) {my $foo = Fcntl​::F_GETFL()};'
runs

The warning message happens in the BEGIN block for sure. if I take -w out, the code does not fail.

I think it's related to this comment in Fcntl.pm but I'm not sure.
# Force the constants to become inlined
BEGIN {
XSLoader​::load 'Fcntl', $VERSION;
}

Thanks for the report.

It seems to be completely unrelated to Fnctl or any external module​:

$ ./perl -we 'sub A () {1}; if (0) {my $foo = A}'
$ ./perl -we 'sub A () {1}; if (0) {my $foo = A or die}'
Found = in conditional, should be == at -e line 1.
$

It seems to be unrelated to constant folding​:

$ ./perl -cwe 'sub A () {1}; {my $foo = A or die}'
Found = in conditional, should be == at -e line 1.
-e syntax OK
$ ./perl -cwe 'sub A () {1}; {my $foo = A}'
-e syntax OK

Or to subroutines that inline as constants​:

$ ./perl -we 'my $foo = 1'
$ ./perl -we 'my $foo = 1 or die'
Found = in conditional, should be == at -e line 1.
$

I think it's because the construction is equivalent to an if​:

$ ./perl -Ilib -MO=Deparse -we 'my $foo = 1 or die'
Found = in conditional, should be == at -e line 1.
BEGIN { $^W = 1; }
die unless my $foo = 1;
-e syntax OK

So, if the warning on if is desired, I'm not sure if this is a bug.

Nicholas Clark

[p5pRT]

From ebhanssen@cpan.org

On Sat, Sep 11, 2010 at 11​:29 AM, Nicholas Clark <nick@​ccl4.org> wrote​:

On Thu, Sep 09, 2010 at 12​:15​:24PM -0700, Todd Rinaldo wrote​:

This breaks​:
$>perl5.12.1 -w -e 'use Fcntl; print "never runs"; if(0) {my $foo =
Fcntl​::F_GETFL() or die};'
Found = in conditional, should be == at -e line 1.

  <snip>

So, if the warning on if is desired, I'm not sure if this is a bug.

  It's not the warning that's the bug. It's the termination during
compilation, as demonstrated by "never runs" not printing.

  Observe​:

eirik@​bluebird[11​:44​:23]~$ perl5.10.1 -wle 'use Fcntl; print "used to run";
if(0) {my $foo = Fcntl​::F_GETFL() or die};'
Found = in conditional, should be == at -e line 1.
used to run

Eirik

[p5pRT]

From @nwc10

On Sat, Sep 11, 2010 at 11​:46​:48AM +0200, Eirik Berg Hanssen wrote​:

On Sat, Sep 11, 2010 at 11​:29 AM, Nicholas Clark <nick@​ccl4.org> wrote​:

On Thu, Sep 09, 2010 at 12​:15​:24PM -0700, Todd Rinaldo wrote​:

This breaks​:
$>perl5.12.1 -w -e 'use Fcntl; print "never runs"; if(0) {my $foo =
Fcntl​::F_GETFL() or die};'
Found = in conditional, should be == at -e line 1.

<snip>

So, if the warning on if is desired, I'm not sure if this is a bug.

It's not the warning that's the bug. It's the termination during
compilation, as demonstrated by "never runs" not printing.

Observe​:

eirik@​bluebird[11​:44​:23]~$ perl5.10.1 -wle 'use Fcntl; print "used to run";
if(0) {my $foo = Fcntl​::F_GETFL() or die};'
Found = in conditional, should be == at -e line 1.
used to run

I cannot reproduce this. This is 5.12.1​:

$ LD_LIBRARY_PATH=/home/nick/Perl/perl ./perl -Ilib -wle 'use Fcntl; print "used to run"; if(0) {my $foo = Fcntl​::F_GETFL() or die};'
Found = in conditional, should be == at -e line 1.
used to run
$ LD_LIBRARY_PATH=/home/nick/Perl/perl ./perl -Ilib -VSummary of my perl5 (revision 5 version 12 subversion 1) configuration​:
  Derived from​: b85be81
  Platform​:
  osname=linux, osvers=2.6.18.8-xenu, archname=x86_64-linux
  uname='linux eris 2.6.18.8-xenu #1 smp sat oct 3 10​:27​:42 bst 2009 x86_64 gnulinux '
  config_args='-Dusedevel=y -Dcc=ccache gcc -Dld=gcc -Ubincompat5005 -Uinstallusrbinperl -Dcf_email=nick@​ccl4.org -Dperladmin=nick@​ccl4.org -Dinc_version_list= -Dinc_version_list_init=0 -Doptimize=-Os -Uusethreads -Uuse64bitall -Uusemymalloc -Duseperlio -Dprefix=/Sandpit/snap5.9.x-v5.12.1 -Uusevendorprefix -Uvendorprefix=/Sandpit/snap5.9.x-v5.12.1 -Dinstallman1dir=none -Dinstallman3dir=none -Uuserelocatableinc -Umad -Dusemymalloc=y -Duseshrplib=true -de'
  hint=recommended, useposix=true, d_sigaction=define
  useithreads=undef, usemultiplicity=undef
  useperlio=define, d_sfio=undef, uselargefiles=define, usesocks=undef
  use64bitint=define, use64bitall=undef, uselongdouble=undef
  usemymalloc=y, bincompat5005=undef
  Compiler​:
  cc='ccache gcc', ccflags ='-fno-strict-aliasing -pipe -fstack-protector -I/usr/local/include -D_LARGEFILE_SOURCE -D_FILE_OFFSET_BITS=64',
  optimize='-Os',
  cppflags='-fno-strict-aliasing -pipe -fstack-protector -I/usr/local/include'
  ccversion='', gccversion='4.3.2', gccosandvers=''
  intsize=4, longsize=8, ptrsize=8, doublesize=8, byteorder=12345678
  d_longlong=define, longlongsize=8, d_longdbl=define, longdblsize=16
  ivtype='long', ivsize=8, nvtype='double', nvsize=8, Off_t='off_t', lseeksize=8
  alignbytes=8, prototype=define
  Linker and Libraries​:
  ld='gcc', ldflags =' -fstack-protector -L/usr/local/lib'
  libpth=/usr/local/lib /lib /usr/lib /lib64 /usr/lib64
  libs=-lnsl -ldb -ldl -lm -lcrypt -lutil -lc
  perllibs=-lnsl -ldl -lm -lcrypt -lutil -lc
  libc=/lib/libc-2.7.so, so=so, useshrplib=true, libperl=libperl.so
  gnulibc_version='2.7'
  Dynamic Linking​:
  dlsrc=dl_dlopen.xs, dlext=so, d_dlsymun=undef, ccdlflags='-Wl,-E -Wl,-rpath,/home/nick/Sandpit/snap5.9.x-v5.12.1/lib/perl5/5.12.1/x86_64-linux/CORE'
  cccdlflags='-fPIC', lddlflags='-shared -Os -L/usr/local/lib -fstack-protector'

Characteristics of this binary (from libperl)​:
  Compile-time options​: MYMALLOC PERL_DONT_CREATE_GVSV PERL_MALLOC_WRAP
  PERL_USE_DEVEL USE_64_BIT_INT USE_LARGE_FILES
  USE_PERLIO USE_PERL_ATOF
  Locally applied patches​:
  uncommitted-changes
  Built under linux
  Compiled at Sep 12 2010 10​:15​:20
  @​INC​:
  lib
  /home/nick/Sandpit/snap5.9.x-v5.12.1/lib/perl5/site_perl/5.12.1/x86_64-linux
  /home/nick/Sandpit/snap5.9.x-v5.12.1/lib/perl5/site_perl/5.12.1
  /home/nick/Sandpit/snap5.9.x-v5.12.1/lib/perl5/5.12.1/x86_64-linux
  /home/nick/Sandpit/snap5.9.x-v5.12.1/lib/perl5/5.12.1
  .

Nicholas Clark

[p5pRT]

From @toddr

On Sep 12, 2010, at 4​:23 AM, Nicholas Clark wrote​:

On Sat, Sep 11, 2010 at 11​:46​:48AM +0200, Eirik Berg Hanssen wrote​:

On Sat, Sep 11, 2010 at 11​:29 AM, Nicholas Clark <nick@​ccl4.org> wrote​:

On Thu, Sep 09, 2010 at 12​:15​:24PM -0700, Todd Rinaldo wrote​:

This breaks​:
$>perl5.12.1 -w -e 'use Fcntl; print "never runs"; if(0) {my $foo =
Fcntl​::F_GETFL() or die};'
Found = in conditional, should be == at -e line 1.

I cannot reproduce this. This is 5.12.1​:

$ LD_LIBRARY_PATH=/home/nick/Perl/perl ./perl -Ilib -wle 'use Fcntl; print "used to run"; if(0) {my $foo = Fcntl​::F_GETFL() or die};'
Found = in conditional, should be == at -e line 1.
used to run

My bad. My shell was chopping off the never runs string because it was missing a newline in the print. This part of the example is not an issue.

Todd Rinaldo

[p5pRT]

From @toddr

On Sep 11, 2010, at 4​:29 AM, Nicholas Clark wrote​:

Thanks for the report.

It seems to be completely unrelated to Fnctl or any external module​:

$ ./perl -we 'sub A () {1}; if (0) {my $foo = A}'
$ ./perl -we 'sub A () {1}; if (0) {my $foo = A or die}'
Found = in conditional, should be == at -e line 1.
$

It seems to be unrelated to constant folding​:

$ ./perl -cwe 'sub A () {1}; {my $foo = A or die}'
Found = in conditional, should be == at -e line 1.
-e syntax OK
$ ./perl -cwe 'sub A () {1}; {my $foo = A}'
-e syntax OK

Or to subroutines that inline as constants​:

$ ./perl -we 'my $foo = 1'
$ ./perl -we 'my $foo = 1 or die'
Found = in conditional, should be == at -e line 1.
$

I think it's because the construction is equivalent to an if​:

$ ./perl -Ilib -MO=Deparse -we 'my $foo = 1 or die'
Found = in conditional, should be == at -e line 1.
BEGIN { $^W = 1; }
die unless my $foo = 1;
-e syntax OK

So, if the warning on if is desired, I'm not sure if this is a bug.

Nicholas Clark

Nicholas, thanks for the reply. To rephrase​:

As I understand it,

This warns because putting an or next to a constant assignment is silly and potentially meant to be ==​:
$ ./perl -we 'my $foo = 1 or die'
Found = in conditional, should be == at -e line 1.

So the issue is that Fcntl is inlining the subroutine calls into constants, which makes

my $foo = Fcntl​::F_GETFL() or die;

inline to

my $foo = 3 or die;

The code in question that brought this up needed to be cross platform. I wasn't certain if Fcntl sub calls would always give sane answers for these constants and was checking their values. I guess I just need to trust that it will always return a valid constant for the given platform?

Thanks,
Todd

[p5pRT]

From @cpansprout

On Mon Sep 20 12​:33​:23 2010, toddr@​cpanel.net wrote​:

On Sep 11, 2010, at 4​:29 AM, Nicholas Clark wrote​:

Thanks for the report.

It seems to be completely unrelated to Fnctl or any external module​:

$ ./perl -we 'sub A () {1}; if (0) {my $foo = A}'
$ ./perl -we 'sub A () {1}; if (0) {my $foo = A or die}'
Found = in conditional, should be == at -e line 1.
$

Nicholas, thanks for the reply. To rephrase​:

As I understand it,

This warns because putting an or next to a constant assignment is
silly and potentially meant to be ==​:
$ ./perl -we 'my $foo = 1 or die'
Found = in conditional, should be == at -e line 1.

So the issue is that Fcntl is inlining the subroutine calls into
constants, which makes

my $foo = Fcntl​::F_GETFL() or die;

inline to

my $foo = 3 or die;

The code in question that brought this up needed to be cross platform.
I wasn't certain if Fcntl sub calls would always give sane answers
for these constants and was checking their values. I guess I just
need to trust that it will always return a valid constant for the
given platform?

With this four-line patch​:

$ ./perl -we 'sub A () {1}; if (0) {my $foo = A or die}'
$ ./perl -we 'sub A () {1}; if (0) {my $foo = 1 or die}'
Found = in conditional, should be == at -e line 1.

Since the value of a constant may not be known at the time the program
is written, it should be perfectly acceptable to do a constant
assignment in a conditional.

This just needs a test added, which I will do if this change is
acceptable. Is it?

[p5pRT]

From @cpansprout

Inline Patch

--- blead-77762.base/op.c	2010-11-18 07:11:12.000000000 -0800
+++ blead-77762-fcntl/op.c	2010-11-23 05:21:54.000000000 -0800
@@ -910,7 +910,8 @@ S_scalarboolean(pTHX_ OP *o)
 
     PERL_ARGS_ASSERT_SCALARBOOLEAN;
 
-    if (o->op_type == OP_SASSIGN && cBINOPo->op_first->op_type == OP_CONST) {
+    if (o->op_type == OP_SASSIGN && cBINOPo->op_first->op_type == OP_CONST
+     && !(cBINOPo->op_first->op_flags & OPf_SPECIAL)) {
 	if (ckWARN(WARN_SYNTAX)) {
 	    const line_t oldline = CopLINE(PL_curcop);
 
diff -Nup blead-77762.base/op.h blead-77762-fcntl/op.h
--- blead-77762.base/op.h	2010-11-17 01:06:10.000000000 -0800
+++ blead-77762-fcntl/op.h	2010-11-23 04:56:39.000000000 -0800
@@ -145,6 +145,7 @@ Deprecated.  Use C<GIMME_V> instead.
 				    operand of a logical or conditional
 				    that was optimised away, so it should
 				    not be bound via =~ */
+				/*  On OP_CONST, from a constant CV */
 
 /* old names; don't use in new code, but don't break them, either */
 #define OPf_LIST	OPf_WANT_LIST
diff -Nup blead-77762.base/toke.c blead-77762-fcntl/toke.c
--- blead-77762.base/toke.c	2010-11-02 21:33:12.000000000 -0700
+++ blead-77762-fcntl/toke.c	2010-11-23 05:10:08.000000000 -0800
@@ -6509,6 +6509,7 @@ Perl_yylex(pTHX)
 			SvREFCNT_dec(((SVOP*)pl_yylval.opval)->op_sv);
 			((SVOP*)pl_yylval.opval)->op_sv = SvREFCNT_inc_simple(sv);
 			pl_yylval.opval->op_private = 0;
+			pl_yylval.opval->op_flags |= OPf_SPECIAL;
 			TOKEN(WORD);
 		    }
 

[p5pRT]

From @cpansprout

On Sun Nov 28 13​:26​:15 2010, sprout wrote​:

On Mon Sep 20 12​:33​:23 2010, toddr@​cpanel.net wrote​:

On Sep 11, 2010, at 4​:29 AM, Nicholas Clark wrote​:

Thanks for the report.

It seems to be completely unrelated to Fnctl or any external module​:

$ ./perl -we 'sub A () {1}; if (0) {my $foo = A}'
$ ./perl -we 'sub A () {1}; if (0) {my $foo = A or die}'
Found = in conditional, should be == at -e line 1.
$

Nicholas, thanks for the reply. To rephrase​:

As I understand it,

This warns because putting an or next to a constant assignment is
silly and potentially meant to be ==​:
$ ./perl -we 'my $foo = 1 or die'
Found = in conditional, should be == at -e line 1.

So the issue is that Fcntl is inlining the subroutine calls into
constants, which makes

my $foo = Fcntl​::F_GETFL() or die;

inline to

my $foo = 3 or die;

The code in question that brought this up needed to be cross platform.
I wasn't certain if Fcntl sub calls would always give sane answers
for these constants and was checking their values. I guess I just
need to trust that it will always return a valid constant for the
given platform?

With this four-line patch​:

$ ./perl -we 'sub A () {1}; if (0) {my $foo = A or die}'
$ ./perl -we 'sub A () {1}; if (0) {my $foo = 1 or die}'
Found = in conditional, should be == at -e line 1.

Since the value of a constant may not be known at the time the program
is written, it should be perfectly acceptable to do a constant
assignment in a conditional.

Tweaked and applied as 6b7c6d9.

[p5pRT]

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