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