File::Glob fails to set error

[p5pRT]

Migrated from rt.perl.org#51410 (status was 'stalled')

Searchable as RT51410$

[p5pRT]

From @nwc10

Created by @nwc10

Test 6 of ext/File/Glob/t/basic.t fails sometimes for me.
It's a test that creates a directory which is unreadable, then tries to glob
it, to make sure that the error gets returned.

It basically does this​:

$ ./perl -Ilib -MFile​::Glob=​:glob -wle 'mkdir "Pie", 0; my @​a = bsd_glob("Pie/*", GLOB_ERR); print GLOB_ERROR, " >@​a<"; rmdir "Pie"'
0 ><
$ ./perl -Ilib -MFile​::Glob=​:glob -wle 'mkdir "Pie", 0; my @​a = bsd_glob("Pie/*", GLOB_ERR); print GLOB_ERROR, " >@​a<"; rmdir "Pie"'
0 ><
$ ./perl -Ilib -MFile​::Glob=​:glob -wle 'mkdir "Pie", 0; my @​a = bsd_glob("Pie/*", GLOB_ERR); print GLOB_ERROR, " >@​a<"; rmdir "Pie"'
-2 ><
$ ./perl -Ilib -MFile​::Glob=​:glob -wle 'mkdir "Pie", 0; my @​a = bsd_glob("Pie/*", GLOB_ERR); print GLOB_ERROR, " >@​a<"; rmdir "Pie"'
-2 ><
$ ./perl -Ilib -MFile​::Glob=​:glob -wle 'mkdir "Pie", 0; my @​a = bsd_glob("Pie/*", GLOB_ERR); print GLOB_ERROR, " >@​a<"; rmdir "Pie"'
-2 ><
$ ./perl -Ilib -MFile​::Glob=​:glob -wle 'mkdir "Pie", 0; my @​a = bsd_glob("Pie/*", GLOB_ERR); print GLOB_ERROR, " >@​a<"; rmdir "Pie"'
-2 ><
$ ./perl -Ilib -MFile​::Glob=​:glob -wle 'mkdir "Pie", 0; my @​a = bsd_glob("Pie/*", GLOB_ERR); print GLOB_ERROR, " >@​a<"; rmdir "Pie"'
-2 ><
$ ./perl -Ilib -MFile​::Glob=​:glob -wle 'mkdir "Pie", 0; my @​a = bsd_glob("Pie/*", GLOB_ERR); print GLOB_ERROR, " >@​a<"; rmdir "Pie"'
-2 ><
$ ./perl -Ilib -MFile​::Glob=​:glob -wle 'mkdir "Pie", 0; my @​a = bsd_glob("Pie/*", GLOB_ERR); print GLOB_ERROR, " >@​a<"; rmdir "Pie"'
0 ><

Um. Yes. Like that. Where -2 is what it expects, and 0 (no error) is not.

So, is this perl being buggy, or a race condition in the filing system?

Surely if I create a directory, I expect it to exist in the next syscall?

[And if it is OS X that is being buggy, it's case sensitive journalled hfs.
And bloody OS X 10.5 has no decent equivalent of strace/truss/ktrace, so I
can't trivially log the syscalls as non-root, and then diff the output.
That's one thing that 10.4 was better at]

I'm leaning towards bsd_glob.c being buggy (is it up to date?) as I wrote a
C program to create, fail to open and delete a directory, all in a tight loop,
and it works just perfectly.

Nicholas Clark

Perl Info

Flags:
    category=library
    severity=low

Site configuration information for perl 5.11.0:

Configured by nick at Mon Mar  3 12:52:51 GMT 2008.

Summary of my perl5 (revision 5 version 11 subversion 0) configuration:
  Platform:
    osname=darwin, osvers=9.2.0, archname=darwin-thread-multi-2level
    uname='darwin mouse-mill.moo 9.2.0 darwin kernel version 9.2.0: tue feb 5 16:13:22 pst 2008; root:xnu-1228.3.13~1release_i386 i386 '
    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=-g -Dusethreads -Duse64bitall -Uusemymalloc -Duseperlio -Dprefix=~/Sandpit/snap5.11.x-33411 -Dinstallman1dir=none -Dinstallman3dir=none -Uuserelocatableinc -Umad -Ud_dosuid -de'
    hint=recommended, useposix=true, d_sigaction=define
    useithreads=define, usemultiplicity=define
    useperlio=define, d_sfio=undef, uselargefiles=define, usesocks=undef
    use64bitint=define, use64bitall=define, uselongdouble=undef
    usemymalloc=n, bincompat5005=undef
  Compiler:
    cc='ccache gcc', ccflags ='-fno-common -DPERL_DARWIN -no-cpp-precomp -arch x86_64 -DDEBUGGING -pipe -fstack-protector -I/opt/local/include',
    optimize='-g',
    cppflags='-no-cpp-precomp -arch x86_64 -fno-common -DPERL_DARWIN -no-cpp-precomp -arch x86_64 -DDEBUGGING -pipe -fstack-protector -I/opt/local/include'
    ccversion='', gccversion='4.0.1 (Apple Inc. build 5465)', 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='env MACOSX_DEPLOYMENT_TARGET=10.3 cc -arch x86_64', ldflags =' -arch x86_64 -fstack-protector -L/opt/local/lib'
    libpth=/opt/local/lib /usr/lib
    libs=-ldbm -ldl -lm -lutil -lc
    perllibs=-ldl -lm -lutil -lc
    libc=/usr/lib/libc.dylib, so=dylib, useshrplib=false, libperl=libperl.a
    gnulibc_version=''
  Dynamic Linking:
    dlsrc=dl_dlopen.xs, dlext=bundle, d_dlsymun=undef, ccdlflags=' '
    cccdlflags=' ', lddlflags=' -bundle -undefined dynamic_lookup -L/opt/local/lib'

Locally applied patches:
    DEVEL


@INC for perl 5.11.0:
    lib
    /Users/nick/Sandpit/snap5.11.x-33411/lib/perl5/5.11.0/darwin-thread-multi-2level
    /Users/nick/Sandpit/snap5.11.x-33411/lib/perl5/5.11.0
    /Users/nick/Sandpit/snap5.11.x-33411/lib/perl5/site_perl/5.11.0/darwin-thread-multi-2level
    /Users/nick/Sandpit/snap5.11.x-33411/lib/perl5/site_perl/5.11.0
    .


Environment for perl 5.11.0:
    DYLD_LIBRARY_PATH (unset)
    HOME=/Users/nick
    LANG=en_GB
    LANGUAGE (unset)
    LD_LIBRARY_PATH (unset)
    LOGDIR (unset)
    PATH=/opt/local/bin:/opt/local/sbin:/Users/nick/bin:/usr/bin:/bin:/usr/sbin:/sbin:/usr/local/bin:/usr/X11/bin:/usr/local/sbin:/sbin:/usr/sbin
    PERL_BADLANG (unset)
    SHELL=/bin/bash

[p5pRT]

From p5p@spam.wizbit.be

On Tue Mar 04 09​:15​:29 2008, nicholas wrote​:

This is a bug report for perl from nick@​ccl4.org,
generated with the help of perlbug 1.36 running under perl 5.11.0.

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

Test 6 of ext/File/Glob/t/basic.t fails sometimes for me.
It's a test that creates a directory which is unreadable, then tries
to glob
it, to make sure that the error gets returned.

It basically does this​:

$ ./perl -Ilib -MFile​::Glob=​:glob -wle 'mkdir "Pie", 0; my @​a =
bsd_glob("Pie/*", GLOB_ERR); print GLOB_ERROR, " >@​a<"; rmdir
"Pie"'
0 ><
$ ./perl -Ilib -MFile​::Glob=​:glob -wle 'mkdir "Pie", 0; my @​a =
bsd_glob("Pie/*", GLOB_ERR); print GLOB_ERROR, " >@​a<"; rmdir
"Pie"'
0 ><
$ ./perl -Ilib -MFile​::Glob=​:glob -wle 'mkdir "Pie", 0; my @​a =
bsd_glob("Pie/*", GLOB_ERR); print GLOB_ERROR, " >@​a<"; rmdir
"Pie"'
-2 ><

Would it be a good idea to expand the test to try it a number of times?
(so that it can be picked up by the smoke reports if it, occasionanly,
fails on other platforms)

Kind regards,

Bram

[p5pRT]

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

[p5pRT]

From @cpansprout

On Sun Jun 29 02​:41​:54 2008, animator wrote​:

On Tue Mar 04 09​:15​:29 2008, nicholas wrote​:

This is a bug report for perl from nick@​ccl4.org,
generated with the help of perlbug 1.36 running under perl 5.11.0.

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

Test 6 of ext/File/Glob/t/basic.t fails sometimes for me.
It's a test that creates a directory which is unreadable, then tries
to glob
it, to make sure that the error gets returned.

It basically does this​:

$ ./perl -Ilib -MFile​::Glob=​:glob -wle 'mkdir "Pie", 0; my @​a =
bsd_glob("Pie/*", GLOB_ERR); print GLOB_ERROR, " >@​a<"; rmdir
"Pie"'
0 ><
$ ./perl -Ilib -MFile​::Glob=​:glob -wle 'mkdir "Pie", 0; my @​a =
bsd_glob("Pie/*", GLOB_ERR); print GLOB_ERROR, " >@​a<"; rmdir
"Pie"'
0 ><
$ ./perl -Ilib -MFile​::Glob=​:glob -wle 'mkdir "Pie", 0; my @​a =
bsd_glob("Pie/*", GLOB_ERR); print GLOB_ERROR, " >@​a<"; rmdir
"Pie"'
-2 ><

I cannot reproduce this on Snow Leopard with 5.6.2, 5.8.1-4, 5.10.0-1,
5.12.0 or 5.13.5.

Would it be a good idea to expand the test to try it a number of times?
(so that it can be picked up by the smoke reports if it, occasionanly,
fails on other platforms)

Has a test been added?

[p5pRT]

From @dcollinsn

It looks like no one but Nick has ever been able to reproduce this. Are you still seeing this on recent perls on Mac?

--
Respectfully,
Dan Collins

[p5pRT]

@dcollinsn - Status changed from 'open' to 'stalled'

[jkeenan]

From @nwc10

Created by @nwc10

Test 6 of ext/File/Glob/t/basic.t fails sometimes for me.
It's a test that creates a directory which is unreadable, then tries to glob
it, to make sure that the error gets returned.

It basically does this​:

$ ./perl -Ilib -MFile​::Glob=​:glob -wle 'mkdir "Pie", 0; my @​a = bsd_glob("Pie/*", GLOB_ERR); print GLOB_ERROR, " >@​a<"; rmdir "Pie"'
0 ><

[snip]

@nwc10 Nick, have you been able to reproduce this error since your original report?

Which test in the current ext/File-Glob/t/basic.t corresponds to the one that was failing for you twelve years ago?

Is the problem now moot since File::Glob::glob has been yanked as of 5.30?

$ ./perl -Ilib -MFile​::Glob=​:glob -wle 'mkdir "Pie", 0; my @​a = bsd_glob("Pie/", GLOB_ERR); print GLOB_ERROR, " >@​a<"; rmdir "Pie"'
0 ><
$ ./perl -Ilib -MFile​::Glob=​:glob -wle 'mkdir "Pie", 0; my @​a = bsd_glob("Pie/", GLOB_ERR); print GLOB_ERROR, " >@​a<"; rmdir "Pie"'
-2 ><
$ ./perl -Ilib -MFile​::Glob=​:glob -wle 'mkdir "Pie", 0; my @​a = bsd_glob("Pie/", GLOB_ERR); print GLOB_ERROR, " >@​a<"; rmdir "Pie"'
-2 ><
$ ./perl -Ilib -MFile​::Glob=​:glob -wle 'mkdir "Pie", 0; my @​a = bsd_glob("Pie/", GLOB_ERR); print GLOB_ERROR, " >@​a<"; rmdir "Pie"'
-2 ><
$ ./perl -Ilib -MFile​::Glob=​:glob -wle 'mkdir "Pie", 0; my @​a = bsd_glob("Pie/", GLOB_ERR); print GLOB_ERROR, " >@​a<"; rmdir "Pie"'
-2 ><
$ ./perl -Ilib -MFile​::Glob=​:glob -wle 'mkdir "Pie", 0; my @​a = bsd_glob("Pie/", GLOB_ERR); print GLOB_ERROR, " >@​a<"; rmdir "Pie"'
-2 ><
$ ./perl -Ilib -MFile​::Glob=​:glob -wle 'mkdir "Pie", 0; my @​a = bsd_glob("Pie/", GLOB_ERR); print GLOB_ERROR, " >@​a<"; rmdir "Pie"'
-2 ><
$ ./perl -Ilib -MFile​::Glob=​:glob -wle 'mkdir "Pie", 0; my @​a = bsd_glob("Pie/", GLOB_ERR); print GLOB_ERROR, " >@​a<"; rmdir "Pie"'
0 ><

Um. Yes. Like that. Where -2 is what it expects, and 0 (no error) is not.

So, is this perl being buggy, or a race condition in the filing system?

Surely if I create a directory, I expect it to exist in the next syscall?

[And if it is OS X that is being buggy, it's case sensitive journalled hfs.
And bloody OS X 10.5 has no decent equivalent of strace/truss/ktrace, so I
can't trivially log the syscalls as non-root, and then diff the output.
That's one thing that 10.4 was better at]

I'm leaning towards bsd_glob.c being buggy (is it up to date?) as I wrote a
C program to create, fail to open and delete a directory, all in a tight loop,
and it works just perfectly.

Thank you very much.
Jim Keenan

[kevinxie4c]

I can reproduce this on CentOS Linux release 7.8.2003 (Core). I got some failed tests when I tried to build Perl5-32.1. When I run

./perl -Ilib -MFile::Glob=:glob -wle 'mkdir "pie", 0; my @a = bsd_glob("pie/*", GLOB_ERR); print GLOB_ERROR,"\n"; rmdir "pie"'

I got "0" all the time when the test expects a non-zero.

Best regards,
Kevin